josephkj / owod Goto Github PK

(CVPR 2021 Oral) Open World Object Detection

License: Apache License 2.0

Python 90.61% Shell 0.84% C++ 3.66% Cuda 4.79% Dockerfile 0.07% CMake 0.03%

open-world object-detection contrastive-learning energy-based-model detectron2 incremental-learning continual-learning cvpr2021 cvpr

owod's Introduction

Towards Open World Object Detection [arXiv | video | poster]

Presented at CVPR 2021 as an ORAL paper

The figure shows how our newly formulated Open World Object Detection setting relates to exsiting settings.

Abstract

Humans have a natural instinct to identify unknown object instances in their environments. The intrinsic curiosity about these unknown instances aids in learning about them, when the corresponding knowledge is eventually available. This motivates us to propose a novel computer vision problem called: Open World Object Detection, where a model is tasked to:

Identify objects that have not been introduced to it as `unknown', without explicit supervision to do so, and
Incrementally learn these identified unknown categories without forgetting previously learned classes, when the corresponding labels are progressively received.

We formulate the problem, introduce a strong evaluation protocol and provide a novel solution, which we call ORE: Open World Object Detector, based on contrastive clustering and energy based unknown identification. Our experimental evaluation and ablation studies analyse the efficacy of ORE in achieving Open World objectives. As an interesting by-product, we find that identifying and characterising unknown instances helps to reduce confusion in an incremental object detection setting, where we achieve state-of-the-art performance, with no extra methodological effort. We hope that our work will attract further research into this newly identified, yet crucial research direction.

A sample qualitative result

The sub-figure (a) is the result produced by our method after learning a few set of classes which doesnot include classes like apple and orange. We are able to identify them and correctly labels them as unknown. After some time, when the model is eventually taught to detect apple and orange, these instances are labelled correctly as seen in sub-figure (b); without forgetting how to detect person. An unidentified class instance still remains, and is successfully detected as an unknown.

Installation

See INSTALL.md.

Dataset setup: Follow these instructions.

Quick Start

Some bookkeeping needs to be done for the code, like removing the local paths and so on. We will update these shortly.

Data split and trained models: [Google Drive Link]

All config files can be found in: configs/OWOD

Sample command on a 4 GPU machine:

python tools/train_net.py --num-gpus 4 --config-file <Change to the appropriate config file> SOLVER.IMS_PER_BATCH 4 SOLVER.BASE_LR 0.005

Kindly run replicate.sh to replicate results from the models shared on the Google Drive.

Kindly check run.sh file for a task workflow.

Incremental Object Detection

If you are interested in Incremental Object Detection, you may have to consider this recent work. We have released the exact training configurations, trained models and logs for all three incremental settings: https://github.com/JosephKJ/iOD

Acknowledgement

Our code base is build on top of Detectron 2 library.

Citation

If you use our work in your research please cite us:

@inproceedings{joseph2021open,
  title={Towards Open World Object Detection},
  author={K J Joseph and Salman Khan and Fahad Shahbaz Khan and Vineeth N Balasubramanian},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2021)},
  eprint={2103.02603},
  archivePrefix={arXiv},
  year={2021}
}

owod's People

Contributors

Stargazers

Watchers

Forkers

sinason rehman04 peternara menglongyue chaoso yangydeng zebrajack yuanmengzhixing oreo-lp abiraja2004 lg12170226 edwardnguyen1705 adamzdw hodapp87 wolfworld6 baodijun joordamn wangzhupi wx-b collector-m peterzs tonystark-001 gaooooke cv-ip gpsbird tlwzzy tianxiajianghu ppzppz yuxiaowei2001 harrylin-hyl xrosliang dannielwang yangbinb shao15xiang shuoyang-1998 big-data-ai vibashan mmaaz60 rush2406 doterkuile draymondliu lvzongyao lab1055 wz940216 czz1240905786 demidovd98 fan945327815 guitaryourself uzair789 jiyuanfeng wenguoquan jessi418 liulei9527 jie311 cvlinks gnehz yidfeng wpfhtl mzq308734881 shining-love xustcr ashishrmenon haleqiu lyk125 doudouwenhui bruinxiong elena-ssq yefeizhu munachisonwadike matteot9890 davelearn summer2474 chandan-iiti thn-buaa jichengyuan hnqin-xdu tianzhengg qddse wukaiyeah ranchogoose yangdesheng shathe daitranskku bigyellowpie wrlife stjordanis jiyang-zheng joeyangrl kartikeya013 outbreak-hui bcsefercik luckychay magicgj vinayasathyanarayana chester-w-xie martinaprogrammer mengqidyangge mama-chengcheng mathpopo 871062304

owod's Issues

Is there a trained model available?

Is there a trained model available at all? It would be great to take a look at.

Thank you very much.

ValueError: All failure and censoring times must be greater than zero.

Hello, dear author
When I use train.yaml to train the model and use val.yaml to train the EBUI component based on the verification set, the following error is reported in the Fit_Weibull_3P:

ValueError: All failure and censoring times must be greater than zero.

I printed my unk and known variables and found that they did have values less than zero.
As I am not familiar with Helmholtz free energy formulation, I hope to get your help. Thank you!

Pretrained models are difficult to download

hi! It's a great job. so i want to start a quick test. Is there a pre-training model quick download link？The data of the link you provided is too large（more than 39G）!
In addition, I noticed that the energy distribution of the set needs to be verified in the test code. Does this mean that we still need a specific data set, rather than being able to detect any open world?

About the Oracle setting in open world object detection task

@JosephKJ Thank you for your great work.

Can you tell me whether the tx_train_with_ukn.yml config is the setting of Oracle in Tab. 2? Does it need other datasets than tx_voc_coco_2007_val_with_known_and_unk?

Using these configs, I can not reproduce the results.
The results in the paper:

and reproduced:

Problem about training schedule

Hello @JosephKJ,
I am confused about the T1~T4 training schedule. There are many training commands in the run.sh, and I have tried to run some experiments. However, I still do not understand the meaning of YAML filenames and which YAML file I should use in each task. Can you release a shell script for reproduction?
Thank you very much!

About The Weibull distribution usage in the code

Hi, I cannot find in the code where the Weibull distribution param is used to predict the unkown.Can you point out that for me?thx

Share file download exceeded quota

hi @JosephKJ , I tried to download the file you shared on GD, but it kept reminding me that I had exceeded the download limit and could not download it. Besides, I tried many ways but failed. Could you please share the data again?Or some other way to share it with me.Thank you very much!

iOD

Hi @JosephKJ ,I wanna train an incremental objects detection model . I want to know which YAML file i should use step by step.

I have trained a model in the execution order - \iOD\19_p_1\base_19_train.yaml -> \iOD\19_p_1\next_1_train_with_ud.yaml.
But it can't detect objects in PREV_INTRODUCED_CLS .

THANKS!

Issue with Reproduction -- Awaiting Response

Hi authors,
I failed to re-produce the reported ORE performance. I couldn't even reproduce the results of faster-rcnn + fine-tuning baseline. In task 3, I got 32 and 11 for previous and current ap50. Compared to the results in your paper, the t3 prev and current is 37 and 12 respectively. The t3 log.txt t3_val_test_log.txt is attached, could you kindly help to fix the problem?
Thanks!

What is the default IMS_PER_BATCH setting

Thank you for your amazing work.

Are the configuration files corresponding to 4 GPU and 4 IMS_PER_BATCH ? Or 16 IMS_PER_BATCH in your experiments?

Do I need to change MAX_ITER and STEPS according to IMS_PER_BATCH?

Thank you so much!

How to run code in win10 system?

Thank you very much for your excellent work. I wonder if I can run the code under win10 system? If you can, can you provide installation documents for me?

Problem about Energy Based Unknown Identifier

Hello @JosephKJ , thanks for your source code.
I am puzzled by how the Energy Based Unknown Identifier mentioned in the paper is implemented in the code.
As described in the paper, the energy function is used to replace the classification head of Faster R-CNN (I think), but the classification head I see in the code is a Linear module named self.cls_score in the FastRCNNOutputLayers class. Thanks!

Some questions

1.In the clustering stage, are unknown classes only divided into one class? if so, for different kinds of unknown classes(some of which may have different features), how to reduce the distance between unknown classes?
2.In the contrastive clustering loss function, are prototypes updated in the previous stage labels for the next stage?
3.The goal of contrastive clustering is to reduce the distance within class and increase the distance between classes.

In the formula, otherwise that I understand is to calculate the distance between classes, but only
calculate the loss that the distance between two classes is not large. But if this part is taken into account, it will shorten the distance between a feature in one category and another. How can this be explained?
4.The labels of unknowns are selected top k boxes according to score from the background classes generated by PRN. But what are unknown region proposals? Do they contain all background boxes?
5.In the algorithm COMPUTECLUSTERINGLOSS, instead of using all original dataset to train the whole network, it retains the previous parameters by assigning weights. It is the contribution of these parameters that enables the network to remember how to identify known classes. In my opinion, this kind of knowledge prolongs the memory time of the network. When more and more new classes are introduced, the network will eventually forget the known classes. Can I understand that?

Thank you!

Cannot find t1_ft folder in models_backup folder

After downloading the owod_bakcup.zip provided by you and unzipping it, I get the following folders -

However, in your t1_test.yaml you have weights in a folder name "t1_ft" (following image) which as you can see is not there in the above image

Request your assistance.

The margin and distance type in your implementation

Thanks for your greate job!
I found that the hyper-parameter 'margin' and the 'distance type' in your code is the opposite of your paper report：
In your code:(detectron2/modeling/roi_heads/fast_rcnn.py/FastRCNNOutputLayers)
self.hingeloss = nn.HingeEmbeddingLoss(2) means the margin is 2.

distances = torch.cdist(fg_features, torch.stack(all_means).cuda(), p=self.margin) where in your code the p(self.margin) is 10, which means a 10-Norm

But in your papar:

You said that the distance is Euclidean distance so the p is should be 2 and the margin should be 10.

I suspect you are writing the reverse
Looking forward to your apply.

[Resolved] Discrepancy in t*_ft.txt

Thanks for your great work !

However, I run the released code but the performance is behind the reported performance in the paper. The first row is the reported performance, the second is my performance. I didn't change any code.

Meanwhile, I find the data split in your logs is different from the released data split in the google drive.
For task 2 finetune, the first image is data split in your logs, the second is that in the google drive. They are apparently different.

Looking forwards your reply.

Error while training for task 2 - UnboundLocalError: local variable 'length' referenced before assignment

Hi,

I'm facing the following error while training for task 2 -

Due to this error, I'm unable to start the training for task 2. I would really appreciate if you could help regarding this error.

Reason for the slow response last month

Hi all,

My apologies for the slow response during April 2021. I was seriously hospitalized with COVID pneumonia. I lost 20 % of my lungs to the virus. I am improving now. Please find the discharge summary attached along, if that helps you to understand better.

discharge_summary.pdf

Thanks,
Joseph

Training procedure for my own dataset to detect unknown objects

Hi @JosephKJ ,I've trained a model which can detect known objects for my own dataset,but it can not detect unknown objects . I want to know what training procedure i should follow for my own dataset (which YAML file i should use step by step) to detect unknown objects.
THANKS!

Same START_ITER values in t_train.yam and t_ft.yaml

Hi,

I noticed that the value of START_ITER is the same in all corresponding train and ft configs. For example, it is START_ITER: 18000 in both t2_train.yaml and t2_ft.yaml. As per the script run.sh, the finetuning (t*_ft.yaml) is using the weights generated from the corresponding t*_train.yaml as starting point, so the START_ITER shouldn't be the same in both files. Please guide.

Thanks

About the visualise_detections.py

Clarification regarding the protocol

Thanks for the nice paper. The paper talks about

"setting the classification logits of the unseen classes to a large negative value (v), thus making their contribution to softmax negligible"

while this explains the training methodology difference between knowns and unknowns on the classification head, I was wondering what is the difference for the bounding box regression head and that for RPN.
That is when training the network initially for a 15+5 class problem setting, the labels for the 5 classes are ignored by fixing the logits for the classification problem but how are the bounding box annotations for these 5 classes used both at the localization head and the RPN?

from unknown class to known class

Hello, I would like to ask, how do you turn an unknown class into a known class? Does it need to be labeled manually?

Inquiry about the validation for EBMs

Thank you for your amazing works.

However, I found that the unknown annotations in t1_valare used to fit the distribution and the model is updated using loss. Then, there are two problems:

Is the pre-trained model used for t1_test from t1_train or t1_val.yaml? Due to the model update in val, I think it should come from t1_train.
In the process of validation, we use the unknown and known annotations to fit EBMs, which can be seen as learning a distribution using small-scaled validation with unknown labels? In this case, will the Open-set degenerate into few-shot learning?

Please correct me if I am the wrong~

KeyError: 'Non-existent config key: OWOD'

HI,
I use the provided docker and report the following error, how can I solve it?

Command Line Args: Namespace(config_file='./configs/OWOD/iOD/all_20_train.yaml', dist_url='tcp://127.0.0.1:50177', eval_only=False, machine_rank=0, num_gpus=1, num_machines=1, opts=[], resume=False)
Traceback (most recent call last):
File "tools/train_net.py", line 169, in
args=(args,),
File "/home/appuser/detectron2_repo/detectron2/engine/launch.py", line 62, in launch
main_func(*args)
File "tools/train_net.py", line 132, in main
cfg = setup(args)
File "tools/train_net.py", line 124, in setup
cfg.merge_from_file(args.config_file)
File "/home/appuser/detectron2_repo/detectron2/config/config.py", line 54, in merge_from_file
self.merge_from_other_cfg(loaded_cfg)
File "/home/appuser/.local/lib/python3.6/site-packages/fvcore/common/config.py", line 123, in merge_from_other_cfg
return super().merge_from_other_cfg(cfg_other)
File "/home/appuser/.local/lib/python3.6/site-packages/yacs/config.py", line 217, in merge_from_other_cfg
_merge_a_into_b(cfg_other, self, self, [])
File "/home/appuser/.local/lib/python3.6/site-packages/yacs/config.py", line 491, in _merge_a_into_b
raise KeyError("Non-existent config key: {}".format(full_key))
KeyError: 'Non-existent config key: OWOD'

No unknown target is found when the trained model is used for inference

Hi, Thanks for your excellent work!
When I used the weights downloaded from google drive to infer my custom dataset, I visualized the detection results and found that the first 20 categories of objects were detected, but the next 60 categories of objects and unknown objects were not detected.
I'm not sure if my config file is wrong or if the energy file is missing.
Here's my config file (I have placed my custom dataset in the corresponding directory of voc_coco_2007_test):

OUTPUT_DIR: "./output/t1"
MODEL:
  META_ARCHITECTURE: "GeneralizedRCNN"
  WEIGHTS: "output/t1/model_final.pth"
  RPN:
    PRE_NMS_TOPK_TEST: 6000
    POST_NMS_TOPK_TEST: 1000
  ROI_HEADS:
    #    NUM_CLASSES: 2 # 0~79 Known class; 80 -> Unknown; 81 -> Background.
    NUM_CLASSES: 81
    NAME: "Res5ROIHeads"
    POSITIVE_FRACTION: 0.25
    NMS_THRESH_TEST: 0.2
    SCORE_THRESH_TEST: 0.7
  MASK_ON: False
  RESNETS:
    DEPTH: 50
DATASETS:
  TRAIN: ("voc_coco_2007_test",)
  TEST: ("voc_coco_2007_test",)
SOLVER:
  IMS_PER_BATCH: 6
  BASE_LR: 0.01
  STEPS: (12000, 16000)
  MAX_ITER: 500
TEST:
  EXPECTED_RESULTS: []
VERSION: 2
OWOD:
  PREV_INTRODUCED_CLS: 20
  CUR_INTRODUCED_CLS: 0
  COMPUTE_ENERGY: False
  ENABLE_CLUSTERING: False
  ENERGY_SAVE_PATH: 'energy'
  SKIP_TRAINING_WHILE_EVAL: True
  ENABLE_CLUSTERING: False
  TEMPERATURE: 1.5
  FEATURE_STORE_SAVE_PATH: 'feature_store'
CUDNN_BENCHMARK: True

And my run command:

python tools/train_net.py --num-gpus 1 --config-file ./output/t1/t1_test.yaml --eval-only

When I run the command, the system displays the following message:

Energy distribution is not found at ./output/t1/energy_dist_20.pkl

Thanks!

What does otherwise mean in Equation 1?

The order of config files execution

Hi, Thanks for your excellent work!
I found the config files in ./configs/OWOD/ folder and the data split files in ./datasets/OWOD_imagesets/ are extremely confusing. What does t#_train_with_unk.yaml mean? What does t#_ft.yaml mean? Should I have to run t#_val.yaml before t#_test.yaml? Can you provide a config file execution order if I want to reproduce the ORE results? E.g., t1_train.yaml->t1_val.yaml->t1_test.yaml->t2_train.yaml ......
It would be very appreciated if you can release several run.sh that consists of all training schedules in the paper.

Thanks!

KeyError: 'Non-existent config key: OWOD'

first thank for your excellent job
but an error occurred when i tried to reproduce your code:

Traceback (most recent call last):
File "tools/train_net.py", line 172, in
args=(args,),
File "/media/chen/299D817A2D97AD94/detectron2/detectron2/engine/launch.py", line 62, in launch
main_func(*args)
File "tools/train_net.py", line 134, in main
cfg = setup(args)
File "tools/train_net.py", line 126, in setup
cfg.merge_from_file(args.config_file)
File "/media/chen/299D817A2D97AD94/detectron2/detectron2/config/config.py", line 55, in merge_from_file
self.merge_from_other_cfg(loaded_cfg)
File "/home/chen/.conda/envs/detectron/lib/python3.6/site-packages/fvcore/common/config.py", line 123, in merge_from_other_cfg
return super().merge_from_other_cfg(cfg_other)
File "/home/chen/.conda/envs/detectron/lib/python3.6/site-packages/yacs/config.py", line 217, in merge_from_other_cfg
_merge_a_into_b(cfg_other, self, self, [])
File "/home/chen/.conda/envs/detectron/lib/python3.6/site-packages/yacs/config.py", line 491, in _merge_a_into_b
raise KeyError("Non-existent config key: {}".format(full_key))

KeyError: 'Non-existent config key: OWOD'

i think i have installed detectron2 successful
look forward to your reply

[resolved] Can anyone reproduce the results? (my results attached)

Does anyone successfully reproduce the results? I ran several times but the results are far away from the author's. My results are attached.

Some question about Energy Based Unknown Identifier

Hello! I am very interested in this paper, but I do not quite understand the Energy Based Unknown Identifier. These energy values are calculated from the output of the classifier. What is the purpose of using the energy values to distinguish whether the label is unknown or not? The label of the target is determined by the training set. What is the purpose of the judgment here?

about visualise_detections.py No such file or directory /home/fk1/workspace/OWOD/output/t1_clustering_val/energy_dist_' + str(20) + '.pkl

Hi, Thanks for your excellent work!

How to run visualise_detections.py?

I run the code visualise_detections.py find the file is /home/fk1/workspace/OWOD/output/t1_clustering_val/energy_dist_' + str(20) + '.pkl "" missin?
I want to see an unknown tag in an image?

Has anyone encountered the problem in the screenshot?

Where to put Annotations and JPEGImages data

Cuda error

Dear Authors,

When I run your code, there is an error about cuda.

CUDA error: invalid device function (ROIAlign_forward_cuda at ....)

Unknown classes

Can the energy obtained by the unknown category be effectively clustered? Can cluster unknown categories into the same cluster

the result cannot be reproduced

At percent , we still cannot reproduce the result from the training scheduler except the author, #26 (comment) could't reproduce WI, either. I don't think anyone can reproduce this result.
I think author should try to reproduce from current code and data. If so, I would be grateful!
look forward to your reply!
@JosephKJ @salman-h-khan

Why use the proposal of an unknown class to regress the gt box of a known class

Thanks for your greate job!
I don't know if I understand correctly. In roi_head stage, is it reasonable to use the proposal of the unknown class to regress the gt box of the known class with the lowest IOU

Do iOD experiments use validation set ?

Hi, It is a very interesting paper! I tried to run the 19_p_1 experiment and was a little confused about the training scheme.

I follow the training sequence you mentioned, base_19 -> next_1_train_with_ud -> ft_with_unk, but the result is 63.1% mAP which is much lower than the result presented on the paper. I check the hyperparameters on the YAML files and find that the hyperparameters ENABLE_CLUSTERING and COMPUTE_ENERGY are set to FALSE on all three YAML files. Do I need to change them for ORE method running?

I read the YAML files for COCO experiments. Seems like, after training, you use a validation set to continue to train the model to acquire energy distribution. For the validation set, all the annotations are provided to generate the known and unknown labels. This doesn't satisfy the incremental learning setting since annotations for old and future classes should not be provided. Could you please classify how do you handle this in your experiments?

Also, in the iOD experiments, I do not find the YAML file for validation set training, do you do validation set training on iOD experiments or it is only for COCO experiments?

Thanks!

Problem about custom dataset

Thanks for your excellent work!
I really want to know how to detect or segment unknown classes using a custom dataset. Because I'm new to this kind of problem, can you give me some advice?

Training Time

Dear Authors，

This is an interesting task. How long does your model need to train？

google drive link limited access ?

@JosephKJ
Tks for sharing such a great work !
Excuse me but I cant download anything from google drive link you had provided (limited access).

How to build the environment on RTX3090?

I followed your settings:

python 3.6
cuda 10.2
nvcc 10.2
pytorch 1.6.0

But I can't install detectron2, I searched many many solutions, but can't make it work.

Sometime it seems to install detectron2 successfully, but I got the error just like #10

KeyError: 'Non-existent config key: OWOD

Then I re-build the detectron2 from the repo, but I got:

ValueError: Unknown CUDA arch (8.6) or GPU not supported

I submit an issue in the original detectron2 repo:
facebookresearch/detectron2#3076

And now I just want to know what is your hardware and software environment in detail, I want to locate where is wrong,
I struggle on this for a whole day, please help me, thank you!

About the energy file

Hi JosephKJ. According to your codes, energy distribution is generated in the execution of validation file. But validation is executed on the dataset which contains unknown instances. I think it's not proper to use information of unknown instances. Do I have any misunderstanding?
Thanks.

Able to reproduce mAP and A-OSE, but not WI for task 1

I can't reproduce t1 result in current version. I didn't change any code and data, and use the model_final in t1_clustering_with_save of backup_model to test(t1_val->t1_test). but my Wilderness Impact result is 0.049, and yours is 0.022 in your results folder.

Can anyone tell me why

my log:

[06/10 04:53:20] detectron2 INFO: Rank of current process: 0. World size: 4
[06/10 04:53:21] detectron2 INFO: Environment info:

sys.platform linux
Python 3.6.9 (default, Jul 17 2020, 12:50:27) [GCC 8.4.0]
numpy 1.19.5
detectron2 0.2.1 @/workspace/detectron2
Compiler GCC 7.5
CUDA compiler CUDA 10.1
detectron2 arch flags 6.1
DETECTRON2_ENV_MODULE
PyTorch 1.6.0 @/usr/local/lib/python3.6/dist-packages/torch
PyTorch debug build False
GPU available True
GPU 0,1,2,3,4,5,6,7 GeForce GTX 1080 Ti (arch=6.1)
CUDA_HOME /usr/local/cuda
Pillow 8.2.0
torchvision 0.7.0 @/usr/local/lib/python3.6/dist-packages/torchvision
torchvision arch flags 3.5, 5.0, 6.0, 7.0, 7.5
fvcore 0.1.1.dev200512
cv2 4.4.0

PyTorch built with:

GCC 7.3
C++ Version: 201402
Intel(R) Math Kernel Library Version 2019.0.5 Product Build 20190808 for Intel(R) 64 architecture applications
Intel(R) MKL-DNN v1.5.0 (Git Hash e2ac1fac44c5078ca927cb9b90e1b3066a0b2ed0)
OpenMP 201511 (a.k.a. OpenMP 4.5)
NNPACK is enabled
CPU capability usage: AVX2
CUDA Runtime 10.2
NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75
CuDNN 7.6.5
Magma 2.5.2
Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DUSE_VULKAN_WRAPPER -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_STATIC_DISPATCH=OFF,

[06/10 04:53:21] detectron2 INFO: Command line arguments: Namespace(config_file='./configs/OWOD/t1/t1_val.yaml', dist_url='tcp://127.0.0.1:52133', eval_only=False, machine_rank=0, num_gpus=4, num_machines=1, opts=['SOLVER.IMS_PER_BATCH', '4', 'SOLVER.BASE_LR', '0.01', 'OWOD.TEMPERATURE', '1.5', 'OUTPUT_DIR', './output/t1_final', 'MODEL.WEIGHTS', 'backup/t1_clustering_with_save/model_final.pth'], resume=False)
[06/10 04:53:21] detectron2 INFO: Contents of args.config_file=./configs/OWOD/t1/t1_val.yaml:
BASE: "../../Base-RCNN-C4-OWOD.yaml"
MODEL:
WEIGHTS: "/home/joseph/workspace/OWOD/output/t1_ft/model_final.pth"
DATASETS:
TRAIN: ('voc_coco_2007_val', ) # t1_voc_coco_2007_train, t1_voc_coco_2007_ft
TEST: ('voc_coco_2007_val', ) # voc_coco_2007_test
SOLVER:
STEPS: (12000, 16000)
MAX_ITER: 500
WARMUP_ITERS: 0
OUTPUT_DIR: "./output/temp_3"
OWOD:
PREV_INTRODUCED_CLS: 0
CUR_INTRODUCED_CLS: 20
COMPUTE_ENERGY: True
ENERGY_SAVE_PATH: 'energy'
SKIP_TRAINING_WHILE_EVAL: False
ENABLE_CLUSTERING: False
TEMPERATURE: 1.5
[06/10 04:53:21] detectron2 INFO: Running with full config:
CUDNN_BENCHMARK: False
DATALOADER:
ASPECT_RATIO_GROUPING: True
FILTER_EMPTY_ANNOTATIONS: True
NUM_WORKERS: 4
REPEAT_THRESHOLD: 0.0
SAMPLER_TRAIN: TrainingSampler
DATASETS:
PRECOMPUTED_PROPOSAL_TOPK_TEST: 1000
PRECOMPUTED_PROPOSAL_TOPK_TRAIN: 2000
PROPOSAL_FILES_TEST: ()
PROPOSAL_FILES_TRAIN: ()
TEST: ('voc_coco_2007_val',)
TRAIN: ('voc_coco_2007_val',)
GLOBAL:
HACK: 1.0
INPUT:
CROP:
ENABLED: False
SIZE: [0.9, 0.9]
TYPE: relative_range
FORMAT: BGR
MASK_FORMAT: polygon
MAX_SIZE_TEST: 1333
MAX_SIZE_TRAIN: 1333
MIN_SIZE_TEST: 800
MIN_SIZE_TRAIN: (480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800)
MIN_SIZE_TRAIN_SAMPLING: choice
RANDOM_FLIP: horizontal
MODEL:
ANCHOR_GENERATOR:
ANGLES: [[-90, 0, 90]]
ASPECT_RATIOS: [[0.5, 1.0, 2.0]]
NAME: DefaultAnchorGenerator
OFFSET: 0.0
SIZES: [[32, 64, 128, 256, 512]]
BACKBONE:
FREEZE_AT: 2
NAME: build_resnet_backbone
DEVICE: cuda
FPN:
FUSE_TYPE: sum
IN_FEATURES: []
NORM:
OUT_CHANNELS: 256
KEYPOINT_ON: False
LOAD_PROPOSALS: False
MASK_ON: False
META_ARCHITECTURE: GeneralizedRCNN
PANOPTIC_FPN:
COMBINE:
ENABLED: True
INSTANCES_CONFIDENCE_THRESH: 0.5
OVERLAP_THRESH: 0.5
STUFF_AREA_LIMIT: 4096
INSTANCE_LOSS_WEIGHT: 1.0
PIXEL_MEAN: [103.53, 116.28, 123.675]
PIXEL_STD: [1.0, 1.0, 1.0]
PROPOSAL_GENERATOR:
MIN_SIZE: 0
NAME: RPN
RESNETS:
DEFORM_MODULATED: False
DEFORM_NUM_GROUPS: 1
DEFORM_ON_PER_STAGE: [False, False, False, False]
DEPTH: 50
NORM: FrozenBN
NUM_GROUPS: 1
OUT_FEATURES: ['res4']
RES2_OUT_CHANNELS: 256
RES5_DILATION: 1
STEM_OUT_CHANNELS: 64
STRIDE_IN_1X1: True
WIDTH_PER_GROUP: 64
RETINANET:
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0)
FOCAL_LOSS_ALPHA: 0.25
FOCAL_LOSS_GAMMA: 2.0
IN_FEATURES: ['p3', 'p4', 'p5', 'p6', 'p7']
IOU_LABELS: [0, -1, 1]
IOU_THRESHOLDS: [0.4, 0.5]
NMS_THRESH_TEST: 0.5
NORM:
NUM_CLASSES: 80
NUM_CONVS: 4
PRIOR_PROB: 0.01
SCORE_THRESH_TEST: 0.05
SMOOTH_L1_LOSS_BETA: 0.1
TOPK_CANDIDATES_TEST: 1000
ROI_BOX_CASCADE_HEAD:
BBOX_REG_WEIGHTS: ((10.0, 10.0, 5.0, 5.0), (20.0, 20.0, 10.0, 10.0), (30.0, 30.0, 15.0, 15.0))
IOUS: (0.5, 0.6, 0.7)
ROI_BOX_HEAD:
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_LOSS_WEIGHT: 1.0
BBOX_REG_WEIGHTS: (10.0, 10.0, 5.0, 5.0)
CLS_AGNOSTIC_BBOX_REG: False
CONV_DIM: 256
FC_DIM: 1024
NAME:
NORM:
NUM_CONV: 0
NUM_FC: 0
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
SMOOTH_L1_BETA: 0.0
TRAIN_ON_PRED_BOXES: False
ROI_HEADS:
BATCH_SIZE_PER_IMAGE: 512
IN_FEATURES: ['res4']
IOU_LABELS: [0, 1]
IOU_THRESHOLDS: [0.5]
NAME: Res5ROIHeads
NMS_THRESH_TEST: 0.5
NUM_CLASSES: 81
POSITIVE_FRACTION: 0.25
PROPOSAL_APPEND_GT: True
SCORE_THRESH_TEST: 0.05
ROI_KEYPOINT_HEAD:
CONV_DIMS: (512, 512, 512, 512, 512, 512, 512, 512)
LOSS_WEIGHT: 1.0
MIN_KEYPOINTS_PER_IMAGE: 1
NAME: KRCNNConvDeconvUpsampleHead
NORMALIZE_LOSS_BY_VISIBLE_KEYPOINTS: True
NUM_KEYPOINTS: 17
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
ROI_MASK_HEAD:
CLS_AGNOSTIC_MASK: False
CONV_DIM: 256
NAME: MaskRCNNConvUpsampleHead
NORM:
NUM_CONV: 0
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
RPN:
BATCH_SIZE_PER_IMAGE: 256
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_LOSS_WEIGHT: 1.0
BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0)
BOUNDARY_THRESH: -1
HEAD_NAME: StandardRPNHead
IN_FEATURES: ['res4']
IOU_LABELS: [0, -1, 1]
IOU_THRESHOLDS: [0.3, 0.7]
LOSS_WEIGHT: 1.0
NMS_THRESH: 0.7
POSITIVE_FRACTION: 0.5
POST_NMS_TOPK_TEST: 1000
POST_NMS_TOPK_TRAIN: 2000
PRE_NMS_TOPK_TEST: 6000
PRE_NMS_TOPK_TRAIN: 12000
SMOOTH_L1_BETA: 0.0
SEM_SEG_HEAD:
COMMON_STRIDE: 4
CONVS_DIM: 128
IGNORE_VALUE: 255
IN_FEATURES: ['p2', 'p3', 'p4', 'p5']
LOSS_WEIGHT: 1.0
NAME: SemSegFPNHead
NORM: GN
NUM_CLASSES: 54
WEIGHTS: backup/t1_clustering_with_save/model_final.pth
OUTPUT_DIR: ./output/t1_final
OWOD:
CLUSTERING:
ITEMS_PER_CLASS: 20
MARGIN: 10.0
MOMENTUM: 0.99
START_ITER: 1000
UPDATE_MU_ITER: 3000
Z_DIMENSION: 128
COMPUTE_ENERGY: True
CUR_INTRODUCED_CLS: 20
ENABLE_CLUSTERING: False
ENABLE_THRESHOLD_AUTOLABEL_UNK: True
ENABLE_UNCERTAINITY_AUTOLABEL_UNK: False
ENERGY_SAVE_PATH: energy
FEATURE_STORE_SAVE_PATH: feature_store
NUM_UNK_PER_IMAGE: 1
PREV_INTRODUCED_CLS: 0
SKIP_TRAINING_WHILE_EVAL: False
TEMPERATURE: 1.5
SEED: -1
SOLVER:
BASE_LR: 0.01
BIAS_LR_FACTOR: 1.0
CHECKPOINT_PERIOD: 5000
CLIP_GRADIENTS:
CLIP_TYPE: value
CLIP_VALUE: 1.0
ENABLED: False
NORM_TYPE: 2.0
GAMMA: 0.1
IMS_PER_BATCH: 4
LR_SCHEDULER_NAME: WarmupMultiStepLR
MAX_ITER: 500
MOMENTUM: 0.9
NESTEROV: False
REFERENCE_WORLD_SIZE: 0
STEPS: (12000, 16000)
WARMUP_FACTOR: 0.001
WARMUP_ITERS: 0
WARMUP_METHOD: linear
WEIGHT_DECAY: 0.0001
WEIGHT_DECAY_BIAS: 0.0001
WEIGHT_DECAY_NORM: 0.0
TEST:
AUG:
ENABLED: False
FLIP: True
MAX_SIZE: 4000
MIN_SIZES: (400, 500, 600, 700, 800, 900, 1000, 1100, 1200)
DETECTIONS_PER_IMAGE: 100
EVAL_PERIOD: 0
EXPECTED_RESULTS: []
KEYPOINT_OKS_SIGMAS: []
PRECISE_BN:
ENABLED: False
NUM_ITER: 200
VERSION: 2
VIS_PERIOD: 0
[06/10 04:53:21] detectron2 INFO: Full config saved to ./output/t1_final/config.yaml
[06/10 04:53:21] d2.utils.env INFO: Using a generated random seed 21659271
[06/10 04:53:22] d2.modeling.roi_heads.fast_rcnn INFO: Invalid class range: [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79]
[06/10 04:53:22] d2.modeling.roi_heads.fast_rcnn INFO: Feature store not found in ./output/t1_final/feature_store/feat.pt. Creating new feature store.
[06/10 04:53:22] d2.engine.defaults INFO: Model:
GeneralizedRCNN(
(backbone): ResNet(
(stem): BasicStem(
(conv1): Conv2d(
3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
)
(res2): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv1): Conv2d(
64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
)
(res3): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv1): Conv2d(
256, 128, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(3): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
)
(res4): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
(conv1): Conv2d(
512, 256, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(3): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(4): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(5): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
)
)
(proposal_generator): RPN(
(rpn_head): StandardRPNHead(
(conv): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(objectness_logits): Conv2d(1024, 15, kernel_size=(1, 1), stride=(1, 1))
(anchor_deltas): Conv2d(1024, 60, kernel_size=(1, 1), stride=(1, 1))
)
(anchor_generator): DefaultAnchorGenerator(
(cell_anchors): BufferList()
)
)
(roi_heads): Res5ROIHeads(
(pooler): ROIPooler(
(level_poolers): ModuleList(
(0): ROIAlign(output_size=(14, 14), spatial_scale=0.0625, sampling_ratio=0, aligned=True)
)
)
(res5): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
(conv1): Conv2d(
1024, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
)
(box_predictor): FastRCNNOutputLayers(
(cls_score): Linear(in_features=2048, out_features=82, bias=True)
(bbox_pred): Linear(in_features=2048, out_features=324, bias=True)
(hingeloss): HingeEmbeddingLoss()
)
)
)
[06/10 04:53:23] d2.data.build INFO: Removed 0 images with no usable annotations. 4000 images left.
[06/10 04:53:23] d2.data.build INFO: Known classes: range(0, 20)
[06/10 04:53:23] d2.data.build INFO: Labelling known instances the corresponding label, and unknown instances as unknown...
[06/10 04:53:23] d2.data.build INFO: Distribution of instances among all 81 categories:
�[36m| category | #instances | category | #instances | category | #instances |
|:-------------:|:-------------|:-------------:|:-------------|:----------:|:-------------|
| aeroplane | 90 | bicycle | 215 | bird | 206 |
| boat | 165 | bottle | 1146 | bus | 157 |
| car | 1218 | cat | 174 | chair | 1531 |
| cow | 88 | diningtable | 693 | dog | 247 |
| horse | 105 | motorbike | 252 | person | 8393 |
| pottedplant | 314 | sheep | 163 | sofa | 258 |
| train | 86 | tvmonitor | 269 | truck | 0 |
| traffic light | 0 | fire hydrant | 0 | stop sign | 0 |
| parking meter | 0 | bench | 0 | elephant | 0 |
| bear | 0 | zebra | 0 | giraffe | 0 |
| backpack | 0 | umbrella | 0 | handbag | 0 |
| tie | 0 | suitcase | 0 | microwave | 0 |
| oven | 0 | toaster | 0 | sink | 0 |
| refrigerator | 0 | frisbee | 0 | skis | 0 |
| snowboard | 0 | sports ball | 0 | kite | 0 |
| baseball bat | 0 | baseball gl.. | 0 | skateboard | 0 |
| surfboard | 0 | tennis racket | 0 | banana | 0 |
| apple | 0 | sandwich | 0 | orange | 0 |
| broccoli | 0 | carrot | 0 | hot dog | 0 |
| pizza | 0 | donut | 0 | cake | 0 |
| bed | 0 | toilet | 0 | laptop | 0 |
| mouse | 0 | remote | 0 | keyboard | 0 |
| cell phone | 0 | book | 0 | clock | 0 |
| vase | 0 | scissors | 0 | teddy bear | 0 |
| hair drier | 0 | toothbrush | 0 | wine glass | 0 |
| cup | 0 | fork | 0 | knife | 0 |
| spoon | 0 | bowl | 0 | unknown | 12841 |
| | | | | | |
| total | 28611 | | | | |�[0m
[06/10 04:53:23] d2.data.build INFO: Number of datapoints: 4000
[06/10 04:53:23] d2.data.common INFO: Serializing 4000 elements to byte tensors and concatenating them all ...
[06/10 04:53:23] d2.data.common INFO: Serialized dataset takes 2.75 MiB
[06/10 04:53:23] d2.data.dataset_mapper INFO: Augmentations used in training: [ResizeShortestEdge(short_edge_length=(480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800), max_size=1333, sample_style='choice'), RandomFlip()]
[06/10 04:53:23] d2.data.build INFO: Using training sampler TrainingSampler
[06/10 04:53:25] fvcore.common.checkpoint INFO: Loading checkpoint from backup/t1_clustering_with_save/model_final.pth
[06/10 04:53:25] d2.engine.train_loop INFO: Starting training from iteration 0
[06/10 04:53:46] d2.utils.events INFO: eta: 0:05:19 iter: 19 total_loss: 1.153 loss_cls: 0.3643 loss_box_reg: 0.4349 loss_clustering: 0 loss_rpn_cls: 0.1931 loss_rpn_loc: 0.1566 time: 0.6630 data_time: 0.3892 lr: 0.01 max_mem: 2594M
[06/10 04:53:59] d2.utils.events INFO: eta: 0:05:08 iter: 39 total_loss: 1.092 loss_cls: 0.3092 loss_box_reg: 0.3235 loss_clustering: 0 loss_rpn_cls: 0.2272 loss_rpn_loc: 0.1761 time: 0.6681 data_time: 0.0038 lr: 0.01 max_mem: 2604M
[06/10 04:54:13] d2.utils.events INFO: eta: 0:04:55 iter: 59 total_loss: 0.992 loss_cls: 0.3279 loss_box_reg: 0.3103 loss_clustering: 0 loss_rpn_cls: 0.1988 loss_rpn_loc: 0.1696 time: 0.6689 data_time: 0.0038 lr: 0.01 max_mem: 2604M
[06/10 04:54:26] d2.utils.events INFO: eta: 0:04:42 iter: 79 total_loss: 1.105 loss_cls: 0.3057 loss_box_reg: 0.4032 loss_clustering: 0 loss_rpn_cls: 0.2088 loss_rpn_loc: 0.2044 time: 0.6695 data_time: 0.0037 lr: 0.01 max_mem: 2604M
[06/10 04:54:40] d2.utils.events INFO: eta: 0:04:28 iter: 99 total_loss: 1.016 loss_cls: 0.3093 loss_box_reg: 0.3395 loss_clustering: 0 loss_rpn_cls: 0.213 loss_rpn_loc: 0.1582 time: 0.6696 data_time: 0.0037 lr: 0.01 max_mem: 2604M
[06/10 04:54:53] d2.utils.events INFO: eta: 0:04:14 iter: 119 total_loss: 0.8104 loss_cls: 0.2401 loss_box_reg: 0.3092 loss_clustering: 0 loss_rpn_cls: 0.1662 loss_rpn_loc: 0.1155 time: 0.6685 data_time: 0.0036 lr: 0.01 max_mem: 2604M
[06/10 04:55:06] d2.utils.events INFO: eta: 0:04:01 iter: 139 total_loss: 0.9382 loss_cls: 0.2849 loss_box_reg: 0.3282 loss_clustering: 0 loss_rpn_cls: 0.1784 loss_rpn_loc: 0.1421 time: 0.6692 data_time: 0.0035 lr: 0.01 max_mem: 2604M
[06/10 04:55:20] d2.utils.events INFO: eta: 0:03:47 iter: 159 total_loss: 1.027 loss_cls: 0.2785 loss_box_reg: 0.3697 loss_clustering: 0 loss_rpn_cls: 0.1812 loss_rpn_loc: 0.1403 time: 0.6693 data_time: 0.0035 lr: 0.01 max_mem: 2604M
[06/10 04:55:33] d2.utils.events INFO: eta: 0:03:34 iter: 179 total_loss: 1.191 loss_cls: 0.3642 loss_box_reg: 0.4159 loss_clustering: 0 loss_rpn_cls: 0.1832 loss_rpn_loc: 0.1689 time: 0.6701 data_time: 0.0035 lr: 0.01 max_mem: 2643M
[06/10 04:55:47] d2.utils.events INFO: eta: 0:03:21 iter: 199 total_loss: 0.8851 loss_cls: 0.2851 loss_box_reg: 0.3336 loss_clustering: 0 loss_rpn_cls: 0.1789 loss_rpn_loc: 0.1395 time: 0.6696 data_time: 0.0035 lr: 0.01 max_mem: 2643M
[06/10 04:56:00] d2.utils.events INFO: eta: 0:03:07 iter: 219 total_loss: 0.9365 loss_cls: 0.2691 loss_box_reg: 0.3085 loss_clustering: 0 loss_rpn_cls: 0.1821 loss_rpn_loc: 0.1269 time: 0.6705 data_time: 0.0042 lr: 0.01 max_mem: 2643M
[06/10 04:56:14] d2.utils.events INFO: eta: 0:02:54 iter: 239 total_loss: 1.031 loss_cls: 0.2904 loss_box_reg: 0.3653 loss_clustering: 0 loss_rpn_cls: 0.1907 loss_rpn_loc: 0.1373 time: 0.6704 data_time: 0.0037 lr: 0.01 max_mem: 2643M
[06/10 04:56:27] d2.utils.events INFO: eta: 0:02:41 iter: 259 total_loss: 0.9676 loss_cls: 0.2808 loss_box_reg: 0.3114 loss_clustering: 0 loss_rpn_cls: 0.1994 loss_rpn_loc: 0.1527 time: 0.6703 data_time: 0.0038 lr: 0.01 max_mem: 2643M
[06/10 04:56:40] d2.utils.events INFO: eta: 0:02:27 iter: 279 total_loss: 1.04 loss_cls: 0.3054 loss_box_reg: 0.361 loss_clustering: 0 loss_rpn_cls: 0.1894 loss_rpn_loc: 0.1673 time: 0.6705 data_time: 0.0042 lr: 0.01 max_mem: 2643M
[06/10 04:56:54] d2.utils.events INFO: eta: 0:02:14 iter: 299 total_loss: 0.9006 loss_cls: 0.2546 loss_box_reg: 0.322 loss_clustering: 0 loss_rpn_cls: 0.1825 loss_rpn_loc: 0.1389 time: 0.6706 data_time: 0.0036 lr: 0.01 max_mem: 2643M
[06/10 04:57:07] d2.utils.events INFO: eta: 0:02:00 iter: 319 total_loss: 0.8975 loss_cls: 0.249 loss_box_reg: 0.3165 loss_clustering: 0 loss_rpn_cls: 0.1702 loss_rpn_loc: 0.1446 time: 0.6708 data_time: 0.0035 lr: 0.01 max_mem: 2643M
[06/10 04:57:21] d2.utils.events INFO: eta: 0:01:47 iter: 339 total_loss: 1.056 loss_cls: 0.2927 loss_box_reg: 0.3793 loss_clustering: 0 loss_rpn_cls: 0.1672 loss_rpn_loc: 0.1588 time: 0.6712 data_time: 0.0039 lr: 0.01 max_mem: 2643M
[06/10 04:57:34] d2.utils.events INFO: eta: 0:01:33 iter: 359 total_loss: 1.058 loss_cls: 0.3289 loss_box_reg: 0.3908 loss_clustering: 0 loss_rpn_cls: 0.1826 loss_rpn_loc: 0.1518 time: 0.6710 data_time: 0.0037 lr: 0.01 max_mem: 2643M
[06/10 04:57:48] d2.utils.events INFO: eta: 0:01:20 iter: 379 total_loss: 0.9792 loss_cls: 0.2812 loss_box_reg: 0.3731 loss_clustering: 0 loss_rpn_cls: 0.1841 loss_rpn_loc: 0.13 time: 0.6709 data_time: 0.0035 lr: 0.01 max_mem: 2643M
[06/10 04:58:01] d2.utils.events INFO: eta: 0:01:07 iter: 399 total_loss: 0.9255 loss_cls: 0.2702 loss_box_reg: 0.3482 loss_clustering: 0 loss_rpn_cls: 0.1869 loss_rpn_loc: 0.1336 time: 0.6710 data_time: 0.0038 lr: 0.01 max_mem: 2643M
[06/10 04:58:15] d2.utils.events INFO: eta: 0:00:53 iter: 419 total_loss: 0.989 loss_cls: 0.3231 loss_box_reg: 0.3238 loss_clustering: 0 loss_rpn_cls: 0.1602 loss_rpn_loc: 0.1629 time: 0.6715 data_time: 0.0037 lr: 0.01 max_mem: 2643M
[06/10 04:58:28] d2.utils.events INFO: eta: 0:00:40 iter: 439 total_loss: 1.009 loss_cls: 0.2952 loss_box_reg: 0.3646 loss_clustering: 0 loss_rpn_cls: 0.19 loss_rpn_loc: 0.1398 time: 0.6716 data_time: 0.0040 lr: 0.01 max_mem: 2643M
[06/10 04:58:42] d2.utils.events INFO: eta: 0:00:26 iter: 459 total_loss: 0.9227 loss_cls: 0.2407 loss_box_reg: 0.2804 loss_clustering: 0 loss_rpn_cls: 0.1958 loss_rpn_loc: 0.1823 time: 0.6716 data_time: 0.0036 lr: 0.01 max_mem: 2643M
[06/10 04:58:55] d2.utils.events INFO: eta: 0:00:13 iter: 479 total_loss: 0.9756 loss_cls: 0.2497 loss_box_reg: 0.3194 loss_clustering: 0 loss_rpn_cls: 0.1942 loss_rpn_loc: 0.1646 time: 0.6716 data_time: 0.0035 lr: 0.01 max_mem: 2643M
[06/10 04:59:09] fvcore.common.checkpoint INFO: Saving checkpoint to ./output/t1_final/model_final.pth
[06/10 04:59:09] d2.utils.events INFO: eta: 0:00:00 iter: 499 total_loss: 0.9712 loss_cls: 0.2922 loss_box_reg: 0.3268 loss_clustering: 0 loss_rpn_cls: 0.1927 loss_rpn_loc: 0.1722 time: 0.6716 data_time: 0.0036 lr: 0.01 max_mem: 2643M
[06/10 04:59:09] d2.engine.train_loop INFO: Going to analyse the energy files...
[06/10 04:59:09] d2.engine.train_loop INFO: Temperature value: 1.5
[06/10 04:59:09] d2.engine.train_loop INFO: Analysing 0 / 2000
[06/10 04:59:20] d2.engine.train_loop INFO: Analysing 100 / 2000
[06/10 04:59:25] d2.engine.train_loop INFO: Analysing 200 / 2000
[06/10 04:59:29] d2.engine.train_loop INFO: Analysing 300 / 2000
[06/10 04:59:33] d2.engine.train_loop INFO: Analysing 400 / 2000
[06/10 04:59:38] d2.engine.train_loop INFO: Analysing 500 / 2000
[06/10 04:59:43] d2.engine.train_loop INFO: Analysing 600 / 2000
[06/10 04:59:47] d2.engine.train_loop INFO: Analysing 700 / 2000
[06/10 04:59:51] d2.engine.train_loop INFO: Analysing 800 / 2000
[06/10 04:59:56] d2.engine.train_loop INFO: Analysing 900 / 2000
[06/10 05:00:00] d2.engine.train_loop INFO: Analysing 1000 / 2000
[06/10 05:00:04] d2.engine.train_loop INFO: Analysing 1100 / 2000
[06/10 05:00:08] d2.engine.train_loop INFO: Analysing 1200 / 2000
[06/10 05:00:12] d2.engine.train_loop INFO: Analysing 1300 / 2000
[06/10 05:00:16] d2.engine.train_loop INFO: Analysing 1400 / 2000
[06/10 05:00:20] d2.engine.train_loop INFO: Analysing 1500 / 2000
[06/10 05:00:24] d2.engine.train_loop INFO: Analysing 1600 / 2000
[06/10 05:00:29] d2.engine.train_loop INFO: Analysing 1700 / 2000
[06/10 05:00:33] d2.engine.train_loop INFO: Analysing 1800 / 2000
[06/10 05:00:37] d2.engine.train_loop INFO: Analysing 1900 / 2000
[06/10 05:00:41] d2.engine.train_loop INFO: len(unk): 48098
[06/10 05:00:41] d2.engine.train_loop INFO: len(known): 74915
[06/10 05:00:41] d2.engine.train_loop INFO: Fitting Weibull distribution...
[06/10 05:00:43] d2.engine.train_loop INFO: --- 1.760291576385498 seconds ---
[06/10 05:00:45] d2.engine.train_loop INFO: --- 2.2597241401672363 seconds ---
[06/10 05:00:45] d2.engine.train_loop INFO: Pickling the parameters to ./output/t1_final/energy_dist_20.pkl
[06/10 05:00:45] d2.engine.train_loop INFO: Plotting the computed energy values...
[06/10 05:00:47] d2.engine.hooks INFO: Overall training speed: 498 iterations in 0:05:34 (0.6716 s / it)
[06/10 05:00:47] d2.engine.hooks INFO: Total training time: 0:07:13 (0:01:38 on hooks)
[06/10 05:03:48] detectron2 INFO: Rank of current process: 0. World size: 4
[06/10 05:03:49] detectron2 INFO: Environment info:

PyTorch built with:

GCC 7.3
C++ Version: 201402
Intel(R) Math Kernel Library Version 2019.0.5 Product Build 20190808 for Intel(R) 64 architecture applications
Intel(R) MKL-DNN v1.5.0 (Git Hash e2ac1fac44c5078ca927cb9b90e1b3066a0b2ed0)
OpenMP 201511 (a.k.a. OpenMP 4.5)
NNPACK is enabled
CPU capability usage: AVX2
CUDA Runtime 10.2
NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75
CuDNN 7.6.5
Magma 2.5.2
Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DUSE_VULKAN_WRAPPER -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_STATIC_DISPATCH=OFF,

[06/10 05:03:49] detectron2 INFO: Command line arguments: Namespace(config_file='./configs/OWOD/t1/t1_test.yaml', dist_url='tcp://127.0.0.1:49152', eval_only=True, machine_rank=0, num_gpus=4, num_machines=1, opts=['SOLVER.IMS_PER_BATCH', '4', 'SOLVER.BASE_LR', '0.005', 'OUTPUT_DIR', './output/t1_final', 'MODEL.WEIGHTS', 'backup/t1_clustering_with_save/model_final.pth'], resume=False)
[06/10 05:03:49] detectron2 INFO: Contents of args.config_file=./configs/OWOD/t1/t1_test.yaml:
BASE: "../../Base-RCNN-C4-OWOD.yaml"
MODEL:
WEIGHTS: "/home/joseph/workspace/OWOD/output/t1_ft/model_final.pth"
ROI_HEADS:
NMS_THRESH_TEST: 0.4
TEST:
DETECTIONS_PER_IMAGE: 50
DATASETS:
TRAIN: ('t1_voc_coco_2007_train', ) # t1_voc_coco_2007_train, t1_voc_coco_2007_ft
TEST: ('voc_coco_2007_test', ) # voc_coco_2007_test
SOLVER:
STEPS: (12000, 16000)
MAX_ITER: 18000
WARMUP_ITERS: 100
OUTPUT_DIR: "./output/temp_3"
OWOD:
PREV_INTRODUCED_CLS: 0
CUR_INTRODUCED_CLS: 20
[06/10 05:03:49] detectron2 INFO: Running with full config:
CUDNN_BENCHMARK: False
DATALOADER:
ASPECT_RATIO_GROUPING: True
FILTER_EMPTY_ANNOTATIONS: True
NUM_WORKERS: 4
REPEAT_THRESHOLD: 0.0
SAMPLER_TRAIN: TrainingSampler
DATASETS:
PRECOMPUTED_PROPOSAL_TOPK_TEST: 1000
PRECOMPUTED_PROPOSAL_TOPK_TRAIN: 2000
PROPOSAL_FILES_TEST: ()
PROPOSAL_FILES_TRAIN: ()
TEST: ('voc_coco_2007_test',)
TRAIN: ('t1_voc_coco_2007_train',)
GLOBAL:
HACK: 1.0
INPUT:
CROP:
ENABLED: False
SIZE: [0.9, 0.9]
TYPE: relative_range
FORMAT: BGR
MASK_FORMAT: polygon
MAX_SIZE_TEST: 1333
MAX_SIZE_TRAIN: 1333
MIN_SIZE_TEST: 800
MIN_SIZE_TRAIN: (480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800)
MIN_SIZE_TRAIN_SAMPLING: choice
RANDOM_FLIP: horizontal
MODEL:
ANCHOR_GENERATOR:
ANGLES: [[-90, 0, 90]]
ASPECT_RATIOS: [[0.5, 1.0, 2.0]]
NAME: DefaultAnchorGenerator
OFFSET: 0.0
SIZES: [[32, 64, 128, 256, 512]]
BACKBONE:
FREEZE_AT: 2
NAME: build_resnet_backbone
DEVICE: cuda
FPN:
FUSE_TYPE: sum
IN_FEATURES: []
NORM:
OUT_CHANNELS: 256
KEYPOINT_ON: False
LOAD_PROPOSALS: False
MASK_ON: False
META_ARCHITECTURE: GeneralizedRCNN
PANOPTIC_FPN:
COMBINE:
ENABLED: True
INSTANCES_CONFIDENCE_THRESH: 0.5
OVERLAP_THRESH: 0.5
STUFF_AREA_LIMIT: 4096
INSTANCE_LOSS_WEIGHT: 1.0
PIXEL_MEAN: [103.53, 116.28, 123.675]
PIXEL_STD: [1.0, 1.0, 1.0]
PROPOSAL_GENERATOR:
MIN_SIZE: 0
NAME: RPN
RESNETS:
DEFORM_MODULATED: False
DEFORM_NUM_GROUPS: 1
DEFORM_ON_PER_STAGE: [False, False, False, False]
DEPTH: 50
NORM: FrozenBN
NUM_GROUPS: 1
OUT_FEATURES: ['res4']
RES2_OUT_CHANNELS: 256
RES5_DILATION: 1
STEM_OUT_CHANNELS: 64
STRIDE_IN_1X1: True
WIDTH_PER_GROUP: 64
RETINANET:
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0)
FOCAL_LOSS_ALPHA: 0.25
FOCAL_LOSS_GAMMA: 2.0
IN_FEATURES: ['p3', 'p4', 'p5', 'p6', 'p7']
IOU_LABELS: [0, -1, 1]
IOU_THRESHOLDS: [0.4, 0.5]
NMS_THRESH_TEST: 0.5
NORM:
NUM_CLASSES: 80
NUM_CONVS: 4
PRIOR_PROB: 0.01
SCORE_THRESH_TEST: 0.05
SMOOTH_L1_LOSS_BETA: 0.1
TOPK_CANDIDATES_TEST: 1000
ROI_BOX_CASCADE_HEAD:
BBOX_REG_WEIGHTS: ((10.0, 10.0, 5.0, 5.0), (20.0, 20.0, 10.0, 10.0), (30.0, 30.0, 15.0, 15.0))
IOUS: (0.5, 0.6, 0.7)
ROI_BOX_HEAD:
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_LOSS_WEIGHT: 1.0
BBOX_REG_WEIGHTS: (10.0, 10.0, 5.0, 5.0)
CLS_AGNOSTIC_BBOX_REG: False
CONV_DIM: 256
FC_DIM: 1024
NAME:
NORM:
NUM_CONV: 0
NUM_FC: 0
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
SMOOTH_L1_BETA: 0.0
TRAIN_ON_PRED_BOXES: False
ROI_HEADS:
BATCH_SIZE_PER_IMAGE: 512
IN_FEATURES: ['res4']
IOU_LABELS: [0, 1]
IOU_THRESHOLDS: [0.5]
NAME: Res5ROIHeads
NMS_THRESH_TEST: 0.4
NUM_CLASSES: 81
POSITIVE_FRACTION: 0.25
PROPOSAL_APPEND_GT: True
SCORE_THRESH_TEST: 0.05
ROI_KEYPOINT_HEAD:
CONV_DIMS: (512, 512, 512, 512, 512, 512, 512, 512)
LOSS_WEIGHT: 1.0
MIN_KEYPOINTS_PER_IMAGE: 1
NAME: KRCNNConvDeconvUpsampleHead
NORMALIZE_LOSS_BY_VISIBLE_KEYPOINTS: True
NUM_KEYPOINTS: 17
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
ROI_MASK_HEAD:
CLS_AGNOSTIC_MASK: False
CONV_DIM: 256
NAME: MaskRCNNConvUpsampleHead
NORM:
NUM_CONV: 0
POOLER_RESOLUTION: 14
POOLER_SAMPLING_RATIO: 0
POOLER_TYPE: ROIAlignV2
RPN:
BATCH_SIZE_PER_IMAGE: 256
BBOX_REG_LOSS_TYPE: smooth_l1
BBOX_REG_LOSS_WEIGHT: 1.0
BBOX_REG_WEIGHTS: (1.0, 1.0, 1.0, 1.0)
BOUNDARY_THRESH: -1
HEAD_NAME: StandardRPNHead
IN_FEATURES: ['res4']
IOU_LABELS: [0, -1, 1]
IOU_THRESHOLDS: [0.3, 0.7]
LOSS_WEIGHT: 1.0
NMS_THRESH: 0.7
POSITIVE_FRACTION: 0.5
POST_NMS_TOPK_TEST: 1000
POST_NMS_TOPK_TRAIN: 2000
PRE_NMS_TOPK_TEST: 6000
PRE_NMS_TOPK_TRAIN: 12000
SMOOTH_L1_BETA: 0.0
SEM_SEG_HEAD:
COMMON_STRIDE: 4
CONVS_DIM: 128
IGNORE_VALUE: 255
IN_FEATURES: ['p2', 'p3', 'p4', 'p5']
LOSS_WEIGHT: 1.0
NAME: SemSegFPNHead
NORM: GN
NUM_CLASSES: 54
WEIGHTS: backup/t1_clustering_with_save/model_final.pth
OUTPUT_DIR: ./output/t1_final
OWOD:
CLUSTERING:
ITEMS_PER_CLASS: 20
MARGIN: 10.0
MOMENTUM: 0.99
START_ITER: 1000
UPDATE_MU_ITER: 3000
Z_DIMENSION: 128
COMPUTE_ENERGY: False
CUR_INTRODUCED_CLS: 20
ENABLE_CLUSTERING: True
ENABLE_THRESHOLD_AUTOLABEL_UNK: True
ENABLE_UNCERTAINITY_AUTOLABEL_UNK: False
ENERGY_SAVE_PATH:
FEATURE_STORE_SAVE_PATH: feature_store
NUM_UNK_PER_IMAGE: 1
PREV_INTRODUCED_CLS: 0
SKIP_TRAINING_WHILE_EVAL: False
TEMPERATURE: 1.5
SEED: -1
SOLVER:
BASE_LR: 0.005
BIAS_LR_FACTOR: 1.0
CHECKPOINT_PERIOD: 5000
CLIP_GRADIENTS:
CLIP_TYPE: value
CLIP_VALUE: 1.0
ENABLED: False
NORM_TYPE: 2.0
GAMMA: 0.1
IMS_PER_BATCH: 4
LR_SCHEDULER_NAME: WarmupMultiStepLR
MAX_ITER: 18000
MOMENTUM: 0.9
NESTEROV: False
REFERENCE_WORLD_SIZE: 0
STEPS: (12000, 16000)
WARMUP_FACTOR: 0.001
WARMUP_ITERS: 100
WARMUP_METHOD: linear
WEIGHT_DECAY: 0.0001
WEIGHT_DECAY_BIAS: 0.0001
WEIGHT_DECAY_NORM: 0.0
TEST:
AUG:
ENABLED: False
FLIP: True
MAX_SIZE: 4000
MIN_SIZES: (400, 500, 600, 700, 800, 900, 1000, 1100, 1200)
DETECTIONS_PER_IMAGE: 50
EVAL_PERIOD: 0
EXPECTED_RESULTS: []
KEYPOINT_OKS_SIGMAS: []
PRECISE_BN:
ENABLED: False
NUM_ITER: 200
VERSION: 2
VIS_PERIOD: 0
[06/10 05:03:49] detectron2 INFO: Full config saved to ./output/t1_final/config.yaml
[06/10 05:03:49] d2.utils.env INFO: Using a generated random seed 49800563
[06/10 05:03:50] d2.modeling.roi_heads.fast_rcnn INFO: Invalid class range: [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79]
[06/10 05:03:50] d2.modeling.roi_heads.fast_rcnn INFO: Feature store not found in ./output/t1_final/feature_store/feat.pt. Creating new feature store.
[06/10 05:03:50] d2.engine.defaults INFO: Model:
GeneralizedRCNN(
(backbone): ResNet(
(stem): BasicStem(
(conv1): Conv2d(
3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
)
(res2): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv1): Conv2d(
64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv2): Conv2d(
64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=64, eps=1e-05)
)
(conv3): Conv2d(
64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
)
)
(res3): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv1): Conv2d(
256, 128, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
(3): BottleneckBlock(
(conv1): Conv2d(
512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv2): Conv2d(
128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=128, eps=1e-05)
)
(conv3): Conv2d(
128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
)
)
(res4): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
(conv1): Conv2d(
512, 256, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(3): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(4): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
(5): BottleneckBlock(
(conv1): Conv2d(
1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv2): Conv2d(
256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=256, eps=1e-05)
)
(conv3): Conv2d(
256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=1024, eps=1e-05)
)
)
)
)
(proposal_generator): RPN(
(rpn_head): StandardRPNHead(
(conv): Conv2d(1024, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(objectness_logits): Conv2d(1024, 15, kernel_size=(1, 1), stride=(1, 1))
(anchor_deltas): Conv2d(1024, 60, kernel_size=(1, 1), stride=(1, 1))
)
(anchor_generator): DefaultAnchorGenerator(
(cell_anchors): BufferList()
)
)
(roi_heads): Res5ROIHeads(
(pooler): ROIPooler(
(level_poolers): ModuleList(
(0): ROIAlign(output_size=(14, 14), spatial_scale=0.0625, sampling_ratio=0, aligned=True)
)
)
(res5): Sequential(
(0): BottleneckBlock(
(shortcut): Conv2d(
1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
(conv1): Conv2d(
1024, 512, kernel_size=(1, 1), stride=(2, 2), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
(1): BottleneckBlock(
(conv1): Conv2d(
2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
(2): BottleneckBlock(
(conv1): Conv2d(
2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv2): Conv2d(
512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=512, eps=1e-05)
)
(conv3): Conv2d(
512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False
(norm): FrozenBatchNorm2d(num_features=2048, eps=1e-05)
)
)
)
(box_predictor): FastRCNNOutputLayers(
(cls_score): Linear(in_features=2048, out_features=82, bias=True)
(bbox_pred): Linear(in_features=2048, out_features=324, bias=True)
(hingeloss): HingeEmbeddingLoss()
)
)
)
[06/10 05:03:50] fvcore.common.checkpoint INFO: Loading checkpoint from backup/t1_clustering_with_save/model_final.pth
[06/10 05:03:52] d2.data.build INFO: Known classes: range(0, 20)
[06/10 05:03:52] d2.data.build INFO: Labelling known instances the corresponding label, and unknown instances as unknown...
[06/10 05:03:53] d2.data.build INFO: Distribution of instances among all 81 categories:
�[36m| category | #instances | category | #instances | category | #instances |
|:-------------:|:-------------|:-------------:|:-------------|:----------:|:-------------|
| aeroplane | 361 | bicycle | 700 | bird | 800 |
| boat | 607 | bottle | 2339 | bus | 429 |
| car | 3463 | cat | 522 | chair | 3996 |
| cow | 392 | diningtable | 1477 | dog | 697 |
| horse | 455 | motorbike | 587 | person | 18378 |
| pottedplant | 1043 | sheep | 387 | sofa | 686 |
| train | 385 | tvmonitor | 683 | truck | 0 |
| traffic light | 0 | fire hydrant | 0 | stop sign | 0 |
| parking meter | 0 | bench | 0 | elephant | 0 |
| bear | 0 | zebra | 0 | giraffe | 0 |
| backpack | 0 | umbrella | 0 | handbag | 0 |
| tie | 0 | suitcase | 0 | microwave | 0 |
| oven | 0 | toaster | 0 | sink | 0 |
| refrigerator | 0 | frisbee | 0 | skis | 0 |
| snowboard | 0 | sports ball | 0 | kite | 0 |
| baseball bat | 0 | baseball gl.. | 0 | skateboard | 0 |
| surfboard | 0 | tennis racket | 0 | banana | 0 |
| apple | 0 | sandwich | 0 | orange | 0 |
| broccoli | 0 | carrot | 0 | hot dog | 0 |
| pizza | 0 | donut | 0 | cake | 0 |
| bed | 0 | toilet | 0 | laptop | 0 |
| mouse | 0 | remote | 0 | keyboard | 0 |
| cell phone | 0 | book | 0 | clock | 0 |
| vase | 0 | scissors | 0 | teddy bear | 0 |
| hair drier | 0 | toothbrush | 0 | wine glass | 0 |
| cup | 0 | fork | 0 | knife | 0 |
| spoon | 0 | bowl | 0 | unknown | 23320 |
| | | | | | |
| total | 61707 | | | | |�[0m
[06/10 05:03:53] d2.data.build INFO: Number of datapoints: 10246
[06/10 05:03:53] d2.data.common INFO: Serializing 10246 elements to byte tensors and concatenating them all ...
[06/10 05:03:53] d2.data.common INFO: Serialized dataset takes 6.34 MiB
[06/10 05:03:53] d2.data.dataset_mapper INFO: Augmentations used in training: [ResizeShortestEdge(short_edge_length=(800, 800), max_size=1333, sample_style='choice')]
[06/10 05:03:53] d2.evaluation.pascal_voc_evaluation INFO: Loading energy distribution from ./output/t1_final/energy_dist_20.pkl
[06/10 05:03:53] d2.evaluation.evaluator INFO: Start inference on 2562 images
[06/10 05:04:03] d2.evaluation.evaluator INFO: Inference done 11/2562. 0.1747 s / img. ETA=0:07:47
[06/10 05:04:08] d2.evaluation.evaluator INFO: Inference done 39/2562. 0.1739 s / img. ETA=0:07:40
[06/10 05:04:13] d2.evaluation.evaluator INFO: Inference done 65/2562. 0.1783 s / img. ETA=0:07:50
[06/10 05:04:18] d2.evaluation.evaluator INFO: Inference done 93/2562. 0.1777 s / img. ETA=0:07:45
[06/10 05:04:24] d2.evaluation.evaluator INFO: Inference done 116/2562. 0.1843 s / img. ETA=0:07:59
[06/10 05:04:29] d2.evaluation.evaluator INFO: Inference done 141/2562. 0.1866 s / img. ETA=0:08:00
[06/10 05:04:34] d2.evaluation.evaluator INFO: Inference done 167/2562. 0.1859 s / img. ETA=0:07:53
[06/10 05:04:39] d2.evaluation.evaluator INFO: Inference done 190/2562. 0.1894 s / img. ETA=0:07:56
[06/10 05:04:44] d2.evaluation.evaluator INFO: Inference done 216/2562. 0.1889 s / img. ETA=0:07:49
[06/10 05:04:49] d2.evaluation.evaluator INFO: Inference done 244/2562. 0.1876 s / img. ETA=0:07:39
[06/10 05:04:54] d2.evaluation.evaluator INFO: Inference done 269/2562. 0.1880 s / img. ETA=0:07:35
[06/10 05:05:00] d2.evaluation.evaluator INFO: Inference done 295/2562. 0.1886 s / img. ETA=0:07:31
[06/10 05:05:05] d2.evaluation.evaluator INFO: Inference done 319/2562. 0.1895 s / img. ETA=0:07:28
[06/10 05:05:10] d2.evaluation.evaluator INFO: Inference done 346/2562. 0.1888 s / img. ETA=0:07:21
[06/10 05:05:15] d2.evaluation.evaluator INFO: Inference done 371/2562. 0.1895 s / img. ETA=0:07:17
[06/10 05:05:20] d2.evaluation.evaluator INFO: Inference done 399/2562. 0.1881 s / img. ETA=0:07:08
[06/10 05:05:25] d2.evaluation.evaluator INFO: Inference done 424/2562. 0.1890 s / img. ETA=0:07:05
[06/10 05:05:30] d2.evaluation.evaluator INFO: Inference done 450/2562. 0.1892 s / img. ETA=0:07:00
[06/10 05:05:35] d2.evaluation.evaluator INFO: Inference done 478/2562. 0.1882 s / img. ETA=0:06:52
[06/10 05:05:41] d2.evaluation.evaluator INFO: Inference done 506/2562. 0.1875 s / img. ETA=0:06:45
[06/10 05:05:46] d2.evaluation.evaluator INFO: Inference done 534/2562. 0.1868 s / img. ETA=0:06:38
[06/10 05:05:51] d2.evaluation.evaluator INFO: Inference done 561/2562. 0.1865 s / img. ETA=0:06:32
[06/10 05:05:56] d2.evaluation.evaluator INFO: Inference done 587/2562. 0.1865 s / img. ETA=0:06:27
[06/10 05:06:01] d2.evaluation.evaluator INFO: Inference done 613/2562. 0.1868 s / img. ETA=0:06:22
[06/10 05:06:06] d2.evaluation.evaluator INFO: Inference done 639/2562. 0.1868 s / img. ETA=0:06:17
[06/10 05:06:11] d2.evaluation.evaluator INFO: Inference done 666/2562. 0.1866 s / img. ETA=0:06:11
[06/10 05:06:17] d2.evaluation.evaluator INFO: Inference done 692/2562. 0.1866 s / img. ETA=0:06:06
[06/10 05:06:22] d2.evaluation.evaluator INFO: Inference done 715/2562. 0.1875 s / img. ETA=0:06:03
[06/10 05:06:27] d2.evaluation.evaluator INFO: Inference done 741/2562. 0.1876 s / img. ETA=0:05:58
[06/10 05:06:32] d2.evaluation.evaluator INFO: Inference done 766/2562. 0.1879 s / img. ETA=0:05:54
[06/10 05:06:37] d2.evaluation.evaluator INFO: Inference done 792/2562. 0.1879 s / img. ETA=0:05:49
[06/10 05:06:42] d2.evaluation.evaluator INFO: Inference done 818/2562. 0.1879 s / img. ETA=0:05:44
[06/10 05:06:47] d2.evaluation.evaluator INFO: Inference done 842/2562. 0.1884 s / img. ETA=0:05:40
[06/10 05:06:53] d2.evaluation.evaluator INFO: Inference done 868/2562. 0.1886 s / img. ETA=0:05:35
[06/10 05:06:58] d2.evaluation.evaluator INFO: Inference done 893/2562. 0.1888 s / img. ETA=0:05:30
[06/10 05:07:03] d2.evaluation.evaluator INFO: Inference done 918/2562. 0.1890 s / img. ETA=0:05:25
[06/10 05:07:08] d2.evaluation.evaluator INFO: Inference done 944/2562. 0.1889 s / img. ETA=0:05:20
[06/10 05:07:13] d2.evaluation.evaluator INFO: Inference done 969/2562. 0.1892 s / img. ETA=0:05:16
[06/10 05:07:18] d2.evaluation.evaluator INFO: Inference done 996/2562. 0.1890 s / img. ETA=0:05:10
[06/10 05:07:23] d2.evaluation.evaluator INFO: Inference done 1022/2562. 0.1890 s / img. ETA=0:05:05
[06/10 05:07:29] d2.evaluation.evaluator INFO: Inference done 1050/2562. 0.1887 s / img. ETA=0:04:59
[06/10 05:07:34] d2.evaluation.evaluator INFO: Inference done 1076/2562. 0.1888 s / img. ETA=0:04:54
[06/10 05:07:39] d2.evaluation.evaluator INFO: Inference done 1101/2562. 0.1890 s / img. ETA=0:04:49
[06/10 05:07:44] d2.evaluation.evaluator INFO: Inference done 1129/2562. 0.1886 s / img. ETA=0:04:43
[06/10 05:07:49] d2.evaluation.evaluator INFO: Inference done 1155/2562. 0.1886 s / img. ETA=0:04:38
[06/10 05:07:54] d2.evaluation.evaluator INFO: Inference done 1182/2562. 0.1884 s / img. ETA=0:04:32
[06/10 05:07:59] d2.evaluation.evaluator INFO: Inference done 1206/2562. 0.1886 s / img. ETA=0:04:28
[06/10 05:08:04] d2.evaluation.evaluator INFO: Inference done 1232/2562. 0.1886 s / img. ETA=0:04:22
[06/10 05:08:09] d2.evaluation.evaluator INFO: Inference done 1257/2562. 0.1887 s / img. ETA=0:04:18
[06/10 05:08:14] d2.evaluation.evaluator INFO: Inference done 1282/2562. 0.1887 s / img. ETA=0:04:13
[06/10 05:08:20] d2.evaluation.evaluator INFO: Inference done 1307/2562. 0.1890 s / img. ETA=0:04:08
[06/10 05:08:25] d2.evaluation.evaluator INFO: Inference done 1332/2562. 0.1891 s / img. ETA=0:04:03
[06/10 05:08:30] d2.evaluation.evaluator INFO: Inference done 1355/2562. 0.1897 s / img. ETA=0:03:59
[06/10 05:08:35] d2.evaluation.evaluator INFO: Inference done 1377/2562. 0.1902 s / img. ETA=0:03:56
[06/10 05:08:40] d2.evaluation.evaluator INFO: Inference done 1399/2562. 0.1906 s / img. ETA=0:03:52
[06/10 05:08:45] d2.evaluation.evaluator INFO: Inference done 1427/2562. 0.1903 s / img. ETA=0:03:46
[06/10 05:08:51] d2.evaluation.evaluator INFO: Inference done 1453/2562. 0.1902 s / img. ETA=0:03:41
[06/10 05:08:56] d2.evaluation.evaluator INFO: Inference done 1482/2562. 0.1899 s / img. ETA=0:03:34
[06/10 05:09:01] d2.evaluation.evaluator INFO: Inference done 1507/2562. 0.1899 s / img. ETA=0:03:29
[06/10 05:09:06] d2.evaluation.evaluator INFO: Inference done 1533/2562. 0.1899 s / img. ETA=0:03:24
[06/10 05:09:11] d2.evaluation.evaluator INFO: Inference done 1558/2562. 0.1900 s / img. ETA=0:03:19
[06/10 05:09:16] d2.evaluation.evaluator INFO: Inference done 1587/2562. 0.1895 s / img. ETA=0:03:13
[06/10 05:09:21] d2.evaluation.evaluator INFO: Inference done 1612/2562. 0.1897 s / img. ETA=0:03:08
[06/10 05:09:26] d2.evaluation.evaluator INFO: Inference done 1639/2562. 0.1895 s / img. ETA=0:03:03
[06/10 05:09:31] d2.evaluation.evaluator INFO: Inference done 1664/2562. 0.1895 s / img. ETA=0:02:58
[06/10 05:09:36] d2.evaluation.evaluator INFO: Inference done 1691/2562. 0.1894 s / img. ETA=0:02:52
[06/10 05:09:42] d2.evaluation.evaluator INFO: Inference done 1718/2562. 0.1892 s / img. ETA=0:02:47
[06/10 05:09:47] d2.evaluation.evaluator INFO: Inference done 1742/2562. 0.1896 s / img. ETA=0:02:42
[06/10 05:09:52] d2.evaluation.evaluator INFO: Inference done 1765/2562. 0.1899 s / img. ETA=0:02:38
[06/10 05:09:57] d2.evaluation.evaluator INFO: Inference done 1790/2562. 0.1899 s / img. ETA=0:02:33
[06/10 05:10:02] d2.evaluation.evaluator INFO: Inference done 1815/2562. 0.1900 s / img. ETA=0:02:28
[06/10 05:10:07] d2.evaluation.evaluator INFO: Inference done 1840/2562. 0.1901 s / img. ETA=0:02:23
[06/10 05:10:13] d2.evaluation.evaluator INFO: Inference done 1865/2562. 0.1902 s / img. ETA=0:02:18
[06/10 05:10:18] d2.evaluation.evaluator INFO: Inference done 1893/2562. 0.1899 s / img. ETA=0:02:13
[06/10 05:10:23] d2.evaluation.evaluator INFO: Inference done 1921/2562. 0.1897 s / img. ETA=0:02:07
[06/10 05:10:28] d2.evaluation.evaluator INFO: Inference done 1945/2562. 0.1899 s / img. ETA=0:02:02
[06/10 05:10:33] d2.evaluation.evaluator INFO: Inference done 1973/2562. 0.1897 s / img. ETA=0:01:57
[06/10 05:10:38] d2.evaluation.evaluator INFO: Inference done 1998/2562. 0.1898 s / img. ETA=0:01:52
[06/10 05:10:43] d2.evaluation.evaluator INFO: Inference done 2027/2562. 0.1895 s / img. ETA=0:01:46
[06/10 05:10:49] d2.evaluation.evaluator INFO: Inference done 2052/2562. 0.1896 s / img. ETA=0:01:41
[06/10 05:10:54] d2.evaluation.evaluator INFO: Inference done 2078/2562. 0.1896 s / img. ETA=0:01:36
[06/10 05:10:59] d2.evaluation.evaluator INFO: Inference done 2103/2562. 0.1896 s / img. ETA=0:01:31
[06/10 05:11:04] d2.evaluation.evaluator INFO: Inference done 2127/2562. 0.1898 s / img. ETA=0:01:26
[06/10 05:11:09] d2.evaluation.evaluator INFO: Inference done 2150/2562. 0.1901 s / img. ETA=0:01:22
[06/10 05:11:14] d2.evaluation.evaluator INFO: Inference done 2175/2562. 0.1901 s / img. ETA=0:01:17
[06/10 05:11:19] d2.evaluation.evaluator INFO: Inference done 2203/2562. 0.1899 s / img. ETA=0:01:11
[06/10 05:11:24] d2.evaluation.evaluator INFO: Inference done 2227/2562. 0.1901 s / img. ETA=0:01:06
[06/10 05:11:29] d2.evaluation.evaluator INFO: Inference done 2253/2562. 0.1900 s / img. ETA=0:01:01
[06/10 05:11:34] d2.evaluation.evaluator INFO: Inference done 2280/2562. 0.1899 s / img. ETA=0:00:56
[06/10 05:11:40] d2.evaluation.evaluator INFO: Inference done 2306/2562. 0.1899 s / img. ETA=0:00:50
[06/10 05:11:45] d2.evaluation.evaluator INFO: Inference done 2330/2562. 0.1901 s / img. ETA=0:00:46
[06/10 05:11:50] d2.evaluation.evaluator INFO: Inference done 2355/2562. 0.1901 s / img. ETA=0:00:41
[06/10 05:11:55] d2.evaluation.evaluator INFO: Inference done 2382/2562. 0.1900 s / img. ETA=0:00:35
[06/10 05:12:00] d2.evaluation.evaluator INFO: Inference done 2406/2562. 0.1901 s / img. ETA=0:00:31
[06/10 05:12:05] d2.evaluation.evaluator INFO: Inference done 2433/2562. 0.1901 s / img. ETA=0:00:25
[06/10 05:12:10] d2.evaluation.evaluator INFO: Inference done 2458/2562. 0.1901 s / img. ETA=0:00:20
[06/10 05:12:15] d2.evaluation.evaluator INFO: Inference done 2484/2562. 0.1900 s / img. ETA=0:00:15
[06/10 05:12:20] d2.evaluation.evaluator INFO: Inference done 2507/2562. 0.1902 s / img. ETA=0:00:10
[06/10 05:12:26] d2.evaluation.evaluator INFO: Inference done 2533/2562. 0.1902 s / img. ETA=0:00:05
[06/10 05:12:31] d2.evaluation.evaluator INFO: Inference done 2558/2562. 0.1903 s / img. ETA=0:00:00
[06/10 05:12:32] d2.evaluation.evaluator INFO: Total inference time: 0:08:30.307989 (0.199573 s / img per device, on 4 devices)
[06/10 05:12:32] d2.evaluation.evaluator INFO: Total inference pure compute time: 0:08:06 (0.190421 s / img per device, on 4 devices)
[06/10 05:13:33] d2.evaluation.pascal_voc_evaluation INFO: Evaluating voc_coco_2007_test using 2012 metric. Note that results do not use the official Matlab API.
[06/10 05:13:33] d2.evaluation.pascal_voc_evaluation INFO: aeroplane has 1990 predictions.
[06/10 05:13:35] d2.evaluation.pascal_voc_evaluation INFO: bicycle has 2569 predictions.
[06/10 05:13:36] d2.evaluation.pascal_voc_evaluation INFO: bird has 3305 predictions.
[06/10 05:13:36] d2.evaluation.pascal_voc_evaluation INFO: boat has 4130 predictions.
[06/10 05:13:37] d2.evaluation.pascal_voc_evaluation INFO: bottle has 6446 predictions.
[06/10 05:13:37] d2.evaluation.pascal_voc_evaluation INFO: bus has 2355 predictions.
[06/10 05:13:38] d2.evaluation.pascal_voc_evaluation INFO: car has 9965 predictions.
[06/10 05:13:39] d2.evaluation.pascal_voc_evaluation INFO: cat has 1828 predictions.
[06/10 05:13:39] d2.evaluation.pascal_voc_evaluation INFO: chair has 16460 predictions.
[06/10 05:13:40] d2.evaluation.pascal_voc_evaluation INFO: cow has 2310 predictions.
[06/10 05:13:41] d2.evaluation.pascal_voc_evaluation INFO: diningtable has 5626 predictions.
[06/10 05:13:41] d2.evaluation.pascal_voc_evaluation INFO: dog has 2851 predictions.
[06/10 05:13:42] d2.evaluation.pascal_voc_evaluation INFO: horse has 2374 predictions.
[06/10 05:13:42] d2.evaluation.pascal_voc_evaluation INFO: motorbike has 2080 predictions.
[06/10 05:13:43] d2.evaluation.pascal_voc_evaluation INFO: person has 44151 predictions.
[06/10 05:13:46] d2.evaluation.pascal_voc_evaluation INFO: pottedplant has 6559 predictions.
[06/10 05:13:47] d2.evaluation.pascal_voc_evaluation INFO: sheep has 2258 predictions.
[06/10 05:13:47] d2.evaluation.pascal_voc_evaluation INFO: sofa has 4422 predictions.
[06/10 05:13:48] d2.evaluation.pascal_voc_evaluation INFO: train has 2554 predictions.
[06/10 05:13:48] d2.evaluation.pascal_voc_evaluation INFO: tvmonitor has 3943 predictions.
[06/10 05:13:49] d2.evaluation.pascal_voc_evaluation INFO: truck has 1 predictions.
[06/10 05:13:49] d2.evaluation.pascal_voc_evaluation INFO: traffic light has 1 predictions.
[06/10 05:13:49] d2.evaluation.pascal_voc_evaluation INFO: fire hydrant has 1 predictions.
[06/10 05:13:50] d2.evaluation.pascal_voc_evaluation INFO: stop sign has 1 predictions.
[06/10 05:13:50] d2.evaluation.pascal_voc_evaluation INFO: parking meter has 1 predictions.
[06/10 05:13:50] d2.evaluation.pascal_voc_evaluation INFO: bench has 1 predictions.
[06/10 05:13:50] d2.evaluation.pascal_voc_evaluation INFO: elephant has 1 predictions.
[06/10 05:13:51] d2.evaluation.pascal_voc_evaluation INFO: bear has 1 predictions.
[06/10 05:13:51] d2.evaluation.pascal_voc_evaluation INFO: zebra has 1 predictions.
[06/10 05:13:51] d2.evaluation.pascal_voc_evaluation INFO: giraffe has 1 predictions.
[06/10 05:13:52] d2.evaluation.pascal_voc_evaluation INFO: backpack has 1 predictions.
[06/10 05:13:52] d2.evaluation.pascal_voc_evaluation INFO: umbrella has 1 predictions.
[06/10 05:13:52] d2.evaluation.pascal_voc_evaluation INFO: handbag has 1 predictions.
[06/10 05:13:53] d2.evaluation.pascal_voc_evaluation INFO: tie has 1 predictions.
[06/10 05:13:53] d2.evaluation.pascal_voc_evaluation INFO: suitcase has 1 predictions.
[06/10 05:13:53] d2.evaluation.pascal_voc_evaluation INFO: microwave has 1 predictions.
[06/10 05:13:53] d2.evaluation.pascal_voc_evaluation INFO: oven has 1 predictions.
[06/10 05:13:54] d2.evaluation.pascal_voc_evaluation INFO: toaster has 1 predictions.
[06/10 05:13:54] d2.evaluation.pascal_voc_evaluation INFO: sink has 1 predictions.
[06/10 05:13:54] d2.evaluation.pascal_voc_evaluation INFO: refrigerator has 1 predictions.
[06/10 05:13:55] d2.evaluation.pascal_voc_evaluation INFO: frisbee has 1 predictions.
[06/10 05:13:55] d2.evaluation.pascal_voc_evaluation INFO: skis has 1 predictions.
[06/10 05:13:55] d2.evaluation.pascal_voc_evaluation INFO: snowboard has 1 predictions.
[06/10 05:13:55] d2.evaluation.pascal_voc_evaluation INFO: sports ball has 1 predictions.
[06/10 05:13:56] d2.evaluation.pascal_voc_evaluation INFO: kite has 1 predictions.
[06/10 05:13:56] d2.evaluation.pascal_voc_evaluation INFO: baseball bat has 1 predictions.
[06/10 05:13:56] d2.evaluation.pascal_voc_evaluation INFO: baseball glove has 1 predictions.
[06/10 05:13:57] d2.evaluation.pascal_voc_evaluation INFO: skateboard has 1 predictions.
[06/10 05:13:57] d2.evaluation.pascal_voc_evaluation INFO: surfboard has 1 predictions.
[06/10 05:13:57] d2.evaluation.pascal_voc_evaluation INFO: tennis racket has 1 predictions.
[06/10 05:13:58] d2.evaluation.pascal_voc_evaluation INFO: banana has 1 predictions.
[06/10 05:13:58] d2.evaluation.pascal_voc_evaluation INFO: apple has 1 predictions.
[06/10 05:13:58] d2.evaluation.pascal_voc_evaluation INFO: sandwich has 1 predictions.
[06/10 05:13:59] d2.evaluation.pascal_voc_evaluation INFO: orange has 1 predictions.
[06/10 05:13:59] d2.evaluation.pascal_voc_evaluation INFO: broccoli has 1 predictions.
[06/10 05:13:59] d2.evaluation.pascal_voc_evaluation INFO: carrot has 1 predictions.
[06/10 05:14:00] d2.evaluation.pascal_voc_evaluation INFO: hot dog has 1 predictions.
[06/10 05:14:00] d2.evaluation.pascal_voc_evaluation INFO: pizza has 1 predictions.
[06/10 05:14:00] d2.evaluation.pascal_voc_evaluation INFO: donut has 1 predictions.
[06/10 05:14:01] d2.evaluation.pascal_voc_evaluation INFO: cake has 1 predictions.
[06/10 05:14:01] d2.evaluation.pascal_voc_evaluation INFO: bed has 1 predictions.
[06/10 05:14:01] d2.evaluation.pascal_voc_evaluation INFO: toilet has 1 predictions.
[06/10 05:14:02] d2.evaluation.pascal_voc_evaluation INFO: laptop has 1 predictions.
[06/10 05:14:02] d2.evaluation.pascal_voc_evaluation INFO: mouse has 1 predictions.
[06/10 05:14:02] d2.evaluation.pascal_voc_evaluation INFO: remote has 1 predictions.
[06/10 05:14:02] d2.evaluation.pascal_voc_evaluation INFO: keyboard has 1 predictions.
[06/10 05:14:03] d2.evaluation.pascal_voc_evaluation INFO: cell phone has 1 predictions.
[06/10 05:14:03] d2.evaluation.pascal_voc_evaluation INFO: book has 1 predictions.
[06/10 05:14:03] d2.evaluation.pascal_voc_evaluation INFO: clock has 1 predictions.
[06/10 05:14:04] d2.evaluation.pascal_voc_evaluation INFO: vase has 1 predictions.
[06/10 05:14:04] d2.evaluation.pascal_voc_evaluation INFO: scissors has 1 predictions.
[06/10 05:14:04] d2.evaluation.pascal_voc_evaluation INFO: teddy bear has 1 predictions.
[06/10 05:14:05] d2.evaluation.pascal_voc_evaluation INFO: hair drier has 1 predictions.
[06/10 05:14:05] d2.evaluation.pascal_voc_evaluation INFO: toothbrush has 1 predictions.
[06/10 05:14:05] d2.evaluation.pascal_voc_evaluation INFO: wine glass has 1 predictions.
[06/10 05:14:06] d2.evaluation.pascal_voc_evaluation INFO: cup has 1 predictions.
[06/10 05:14:06] d2.evaluation.pascal_voc_evaluation INFO: fork has 1 predictions.
[06/10 05:14:06] d2.evaluation.pascal_voc_evaluation INFO: knife has 1 predictions.
[06/10 05:14:07] d2.evaluation.pascal_voc_evaluation INFO: spoon has 1 predictions.
[06/10 05:14:07] d2.evaluation.pascal_voc_evaluation INFO: bowl has 1 predictions.
[06/10 05:14:07] d2.evaluation.pascal_voc_evaluation INFO: unknown has 32573 predictions.
[06/10 05:14:09] d2.evaluation.pascal_voc_evaluation INFO: Wilderness Impact: {0.1: {50: 0.015074211502782932}, 0.2: {50: 0.022118742724097785}, 0.3: {50: 0.03910034602076124}, 0.4: {50: 0.05038959819752159}, 0.5: {50: 0.049498067919281956}, 0.6: {50: 0.04874401462169388}, 0.7: {50: 0.04493868016351956}, 0.8: {50: 0.04991488217901622}, 0.9: {50: 0.055984794253412656}}
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: avg_precision: {0.1: {50: 0.060781120363826324}, 0.2: {50: 0.060781120363826324}, 0.3: {50: 0.060781120363826324}, 0.4: {50: 0.060781120363826324}, 0.5: {50: 0.060781120363826324}, 0.6: {50: 0.060781120363826324}, 0.7: {50: 0.060781120363826324}, 0.8: {50: 0.060781120363826324}, 0.9: {50: 0.060781120363826324}}
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Absolute OSE (total_num_unk_det_as_known): {50: 7259.0}
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: total_num_unk 23320
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: ['aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor', 'truck', 'traffic light', 'fire hydrant', 'stop sign', 'parking meter', 'bench', 'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag', 'tie', 'suitcase', 'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'frisbee', 'skis', 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard', 'tennis racket', 'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'bed', 'toilet', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone', 'book', 'clock', 'vase', 'scissors', 'teddy bear', 'hair drier', 'toothbrush', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl', 'unknown']
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: AP50: ['79.5', '56.5', '58.3', '42.9', '24.0', '71.7', '54.4', '80.8', '20.3', '70.0', '16.0', '77.7', '80.3', '66.9', '46.7', '30.6', '67.4', '48.0', '76.4', '56.8', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '1.2']
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Precisions50: ['15.3', '17.1', '14.9', '7.8', '13.5', '14.4', '21.4', '25.5', '8.3', '11.7', '9.6', '21.1', '16.2', '20.6', '26.0', '9.1', '11.8', '9.2', '12.8', '12.5', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '6.1']
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Recall50: ['90.1', '67.3', '71.2', '66.5', '40.1', '86.6', '66.5', '91.0', '39.7', '85.7', '38.7', '90.9', '92.6', '77.7', '64.2', '63.7', '82.7', '73.9', '89.0', '77.6', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '0.0', '8.5']
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Current class AP50: 56.26823138862905
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Current class Precisions50: 14.939950796347716
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Current class Recall50: 72.77744750787848
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Known AP50: 56.26823138862905
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Known Precisions50: 14.939950796347716
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Known Recall50: 72.77744750787848
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Unknown AP50: 1.1656362823677229
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Unknown Precisions50: 6.072514045375004
[06/10 05:14:10] d2.evaluation.pascal_voc_evaluation INFO: Unknown Recall50: 8.481989708404802
[06/10 05:14:10] d2.engine.defaults INFO: Evaluation results for voc_coco_2007_test in csv format:
[06/10 05:14:10] d2.evaluation.testing INFO: copypaste: Task: bbox
[06/10 05:14:10] d2.evaluation.testing INFO: copypaste: AP,AP50
[06/10 05:14:10] d2.evaluation.testing INFO: copypaste: 13.9078,13.9078

Can I know the detailed division of data set？

First of all, thank you for your work. I know that the data set is composed of coco and VOC, which is divided into four parts. Then I want to know whether the final format of the data in these four parts is unified into VOC format? And what is the composition of the whole data set? Thank you for your reply

Bbox supervision signal for unknown class

Are ground-truth bounding boxes for unknown class utilized as supervision signal during training?

iOD: 10-10

Hi @JosephKJ, I want to train an incremental detection model for the 10+10 experiment on the VOC dataset. May I know what is the order for using the YAML files step by step?

I have trained the model in the order - \iOD\10_p_10\base_10_train.yaml -> \iOD\10_p_10\next_10_train_with_unk_detection.yaml. However, the results show catastrophic forgetting.

Would it possible for you to release the run.sh for the experiments on the VOC dataset?

Thank you very much.

What is the order in which I should use the configuration files in the T1 ~ T4 experiments?Looking forward to your reply!!

Problems about ablation experiments and Oracle

Thanks for your great paper! I have some questions about the paper.
1.What does Oracle actually mean? Is it an existing detector or something else? I cannot find any paper about it.
2.Row 4 of ablation experiments(Table 4) shows that only ALU can have a good effect on experiments. How it works without contrastive clustering?

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.