daqingliu / nmtree Goto Github PK

hey,

Thanks for your work. i am trying out your algorithm on colab pro.

after getting the below error, i installed Torch 1.0.0 but doesn't still work.

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xba in position 0: invalid start byte

below is the response after installation of torch 1.0.0

Loader loading data.json: data/feats/refcocog_umd/data_dep.json
word vocab size is 6793
object category size is 80
we have 25799 images.
we have 208960 anns.
we have 49822 refs.
we have 95010 sentences.
loading visual feats ...
^C

what could be the problem?

Hello. I've been trying to reproduce the result of this paper. Could you be more specific about how to extract the MAttNet feature used during training?

The training command in README requires a "matt_res_det_feats.pth" file. But I followed the instructions in MAttNet repository, extracted features using Mask R-CNN, and finally got a bunch of ".h5" files.

Any help would be appreciated~ Thanks in advance!

Hi, when I ran the feature extraction code, some problems came up that were difficult to solve, so can you provide all the features, including ground truth and detected , thank you very much.

Hello, When I train the model, I meet the question.Could you please help me settle down this? Thank you

Hi @daqingliu,

Trying to run eval_det,py using the command,
python tools/eval_det.py --id det_nmtree_01 --log_path log --dataset refcocog --split_by umd

I am getting the following error:
FileNotFoundError: [Errno 2] No such file or directory: 'data/detections/refcocog_umd/res101_coco_minus_refer_notime_dets.json'

Could you please tell from where and how can I get the json?

Hi,
I am trying to reproduce the results. During training using the following command,

python tools/train.py --id det_nmtree_01 --dataset refcocog --split_by umd --grounding_model NMTree --data_file data_dep --batch_size 128 --glove glove.840B.300d_dep --visual_feat_file matt_res_gt_feats.pth

I am getting the following error:

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xba in position 0: invalid start byte

I have downloaded the MAttNet features and done the preprocessing part. All the settings are default.

The requirements installed are as follows:
blis==0.7.4
cached-property==1.5.2
catalogue==2.0.1
certifi==2020.12.5
chardet==4.0.0
click==7.1.2
cymem==2.0.5
en-core-web-sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.0.0/en_core_web_sm-3.0.0-py3-none-any.whl
h5py==3.2.1
idna==2.10
importlib-metadata==3.7.3
Jinja2==2.11.3
MarkupSafe==1.1.1
murmurhash==1.0.5
numpy==1.20.1
packaging==20.9
pathy==0.4.0
Pillow==8.1.2
pkg-resources==0.0.0
preshed==3.0.5
pydantic==1.7.3
pyparsing==2.4.7
requests==2.25.1
six==1.15.0
smart-open==3.0.0
spacy==3.0.5
spacy-legacy==3.0.1
srsly==2.4.0
thinc==8.0.2
torch==1.4.0
torchvision==0.5.0
tqdm==4.59.0
typer==0.3.2
typing-extensions==3.7.4.3
urllib3==1.26.4
wasabi==0.8.2
zipp==3.4.1

Complete console logs and error stacktrace as follows:

Loader loading data.json:  data/feats/refcocog_umd/data_dep.json
word vocab size is  6751
object category size is  80
we have 25799 images.
we have 208960 anns.
we have 49822 refs.
we have 95010 sentences.
loading visual feats ...
Traceback (most recent call last):
  File "tools/train.py", line 263, in <module>
    main(opt)
  File "tools/train.py", line 59, in main
    loader = GtLoader(data_json, visual_feats_dir, opt, data_pth)
  File "/media/disk/user/abhinav/NMTree/tools/../utils/gt_loader.py", line 33, in __init__
    map_location=lambda storage, loc: storage, pickle_module=pickle)
  File "/media/disk/user/abhinav/NMTree/venv/lib/python3.7/site-packages/torch/serialization.py", line 529, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/media/disk/user/abhinav/NMTree/venv/lib/python3.7/site-packages/torch/serialization.py", line 702, in _legacy_load
    result = unpickler.load()
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xba in position 0: invalid start byte

We run the code python tools/save_matt_gt_feats.py --dataset refcoco --splitBy unc to save the features. But we found ``IOError: [Errno 2] No such file or directory: output/refcoco_unc/mrcn_cmr_with_st.json'. How do we get the file mrcn_cmr_with_st.json?

Any help would be appreciated~ Thanks in advance!