my89 / imsitu Goto Github PK

I'm interested in reproducing the results from the Zhao et al 2017, but I'm worried about falling afoul of copyright. What license is this dataset released under? (For reference, COCO is released under CC-BY 4.0).

Thanks!

hi there,

The url for original images is the same as the resized one. Could you provide the correct link? Thanks!

Hi,

This paper mentions that an additional 5 million examples were collected using the Semantic Augmentation technique. Is this dataset released and how would I go about downloading it? It would be super useful for my work on this dataset.

Thanks in advance

I'm trying to run predict and retrain your network.

When i run code to predict, i use command:
python3.7 baseline_crf.py --command predict --output_dir . --image_dir ./of500_images_resized --dataset_dir . --weights_file 0.23782721560846543.model --encoding_file ./baseline_models/baseline_encoder --cnn_type resnet_50

It creates error:

command = predict
Traceback (most recent call last):
  File "baseline_crf.py", line 668, in <module>
    encoder = torch.load(args.encoding_file)
  File "/usr/local/lib/python3.7/dist-packages/torch/serialization.py", line 529, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/usr/local/lib/python3.7/dist-packages/torch/serialization.py", line 692, in _legacy_load
    magic_number = pickle_module.load(f, **pickle_load_args)
_pickle.UnpicklingError: invalid load key, ' '.

Can you give me some advice how to fix it?
I tried to use
python3.6 + pytorch1.0
anhd python3.7 + pytorch1.5

But none of them work.

Hi Mark,

May I know which version of FrameNet you were using for creating the data? Thank you!

Hi, the scripts do not work in pytorch version 0.4 and later. Please fix it if possible.