datasets

[liwell24]

Thank you for the great work of open source！ Could you tell me where to download the dataset ？ Looking forward to your reply~

[gzaraunitn]

Hi, thanks for your issue. The HMDB and UCF datasets are both available for download. You can refer to the TA3N work (page 11) for the 12 classes that overlap between the two datasets and their naming, and to CEVT for the chosen open-set split. As for Epic-Kitchens, you can use the official download scripts: you should download the frames (--frames) for kitchens P01, P08 and P22. Best!

[liwell24]

Thank u for your reply！ I met some problem when i reproduce OUVDA on epic-kitchens, the error message as follows

_K-means: loading training set
  0%|          | 0/234 [00:00<?, ?it/s]
Error executing job with overrides: ['setting=ek12', 'attributes.n_attributes=6', 'attributes.tf_idf_threshold=0.5', 'attributes.matching_threshold=0.5', 'attributes.selection=topk', 'attributes.k_clustering=45', 'attributes.tf_idf_topk_source=5', 'attributes.tf_idf_topk_target=5', 'attributes.final_prompt_length=3', 'loss.target.weight=1.0', 'loss.target.filtering=top_k_confident_samples', 'loss.target.k=20', 'loss.target.k_type=percentage', 'data.batch_size=16', 'solver.type=cosine', 'solver.epochs=20', 'solver.lr=6.e-6', 'logging.comet=True', 'logging.tag=ek12', 'logging.project_name=autolabel']
Traceback (most recent call last):
  File "../train.py", line 218, in main
    training_results = training_step(
  File "/home/lw/autolabel-main/loops/training_step.py", line 71, in training_step
    clustering_results = clustering_method(model, loader["target"], config, device)
  File "/home/lw/autolabel-main/utils/attributes.py", line 34, in k_means
    for batch_idx, (
  File "/home/lw/anaconda3/envs/autolabel/lib/python3.8/site-packages/tqdm/std.py", line 1171, in __iter__
    for obj in iterable:
  File "/home/lw/anaconda3/envs/autolabel/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 521, in __next__
    data = self._next_data()
  File "/home/lw/anaconda3/envs/autolabel/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 561, in _next_data
    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration
  File "/home/lw/anaconda3/envs/autolabel/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/lw/anaconda3/envs/autolabel/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/lw/autolabel-main/data/dataset.py", line 207, in __getitem__
    return self.get(video, label, segment_indices, video_id)
  File "/home/lw/autolabel-main/data/dataset.py", line 229, in get
    seg_imgs = [Image.open(frame_paths[p]).convert("RGB")]
IndexError: list index out of range_

How should I solve this problem？

[liwell24]

I have solved the problem . Thank you ! :)

[Better-byte]

How did you solve it? I also encountered this problem.