ywchao / ho-rcnn Goto Github PK

@ywchao Hi, could you please help me, thank you very much
when I running detection with a trained mode according to https://github.com/ywchao/ho-rcnn#running-detection-with-a-trained-model
./experiments/scripts/test_rcnn_caffenet_ho_pconv_ip1_s/01_person.sh

the terminal output the information as the following:

• set -e
• export PYTHONUNBUFFERED=True
• PYTHONUNBUFFERED=True
• mkdir -p experiments/logs/test_rcnn_caffenet_ho_pconv_ip1_s/
  ++ date +%Y-%m-%d_%H-%M-%S
• LOG=experiments/logs/test_rcnn_caffenet_ho_pconv_ip1_s/01_person.txt.2018-01-23_22-08-48
• exec
  ++ tee -a experiments/logs/test_rcnn_caffenet_ho_pconv_ip1_s/01_person.txt.2018-01-23_22-08-48
• echo Logging output to experiments/logs/test_rcnn_caffenet_ho_pconv_ip1_s/01_person.txt.2018-01-23_22-08-48
  Logging output to experiments/logs/test_rcnn_caffenet_ho_pconv_ip1_s/01_person.txt.2018-01-23_22-08-48
• ./fast-rcnn/tools/test_net.py --gpu 0 --def ./models/rcnn_caffenet_ho_pconv_s/test_ip.prototxt --net ./output/ho_1_s/hico_det_train2015/rcnn_caffenet_pconv_ip_iter_150000.caffemodel --imdb hico_det_test2015 -
  -cfg ./experiments/cfgs/rcnn_ho_ip1_s.yml --oid 1
  usage: test_net.py [-h] [--gpu GPU_ID] [--def PROTOTXT] [--net CAFFEMODEL]
  [--cfg CFG_FILE] [--wait WAIT] [--imdb IMDB_NAME] [--comp]
  [--set ...]
  test_net.py: error: unrecognized arguments: --oid 1

I'm trying to reproduce some results on the HICO-DET dataset, but I don't seem to find the correct split between rare and non-rare classes.

According to the paper, rare hoi classes are those having <5 instances in the training set. Rare classes should be 167 of the 600 hoi classes in total.

However, if I filter the training set using this criterion I find these counts:

count<k    num_hois
 5              100
 6              114
 7              123
 8              129
 9              132
10              138
11              144
12              154
13              159
14              164
15              168
16              174
17              178
18              183
19              187
20              190

I checked your evaluation code and I spotted a <10 in eval_one.m, but even with that threshold, the number 167 doesn't come up.

Can you provide a list of exactly which hoi categories you consider rare and which not?
A format similar to the one you have on your website would be ideal.

Thanks!

hello
when I read your paper:Learning to Detect Human-Object Interactions,I want to learn your code,but I can't find the dataset:pre-computed HOI detection on the HICO-DET test set (2.0G)
could you tell me where to find this dataset?
thank you!

I'm trying to evaluate my results on the hico-det dataset by running the evaluation code provided from the repo. However, I'm not quite sure how to do this. Can you give an explanation on how to do this?

hello
have You the proposals file separately? and in which file, the proposals is generated and the positive and negative data are generated.
thanks

@ywchao This interesting project gave a demo to run detection on the HICO-DET test set using a trained HO-RCNN model.
However, the input of the demo is so complex and not easy to understand.
It seems the input contains a few .mat file in the data folder and cache folder, but what's the use of them? and what is the meaning of the data in them? and how could we make our data (.jpg images and the .txt contains box information) to be the input of ho-rcnn?
Could anyone explain it? Thanks!

Hello,

Thank you for releasing the code. I do not have access to MATLAB. Is there a Python version of the code? I'd like to be able to input an mp4 video and get the HOI interactions output on the video.

Thank you,

I've read your paper Learning to Detect Human-Object Interactions which is very good, but i was wondering if you have tried implementing HO-RCNN with Faster R-CNN or YOLO?

How do you convert the output of (https://github.com/ywchao/hoi-det-ui) to the input of this repo (https://github.com/ywchao/ho-rcnn)?

I wonder how to convert the JSON output to anno.mat as the dataset.