PaddleVideo: Action recognition

[CodeRic28]

Following this source: https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/en/model_zoo/recognition/tsm.md
on UCF-24 and UCF101 datasets.

when running the following command: python -B -m paddle.distributed.launch --gpus="0" --log_dir=log_tsm main.py --validate -c configs/recognition/tsm/tsm_ucf101_frames.yaml

I'm getting this error: too many values to unpack
Error:
Loading res5c.conv2._conv.weight:
Loading res5c.conv2._batch_norm.weight:
Loading res5c.conv2._batch_norm.bias:
Loading res5c.conv2._batch_norm._mean:
100%|██████████| 265/265 [00:02<00:00, 97.20it/s]

Loading res5c.conv2._batch_norm._variance:
_--------------------------FIEL PATH ------------------ data/ucf101/ucf101_train_split_1_rawframes.txt
fin============================= <_io.TextIOWrapper name='data/ucf101/ucf101_train_split_1_rawframes.txt' mode='r' encoding='UTF-8'>
LINE SPLIT================================ /home/coderic/projects/action_paddle/paddle_video/PaddleVideo/data/ucf101/rawframes/ApplyEyeMakeup/v_ApplyEyeMakeup_g08_c01
Traceback (most recent call last):
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/main.py", line 141, in
main()
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/main.py", line 129, in main
train_model(cfg,
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/tasks/train.py", line 111, in train_model
train_dataset = build_dataset((cfg.DATASET.train, cfg.PIPELINE.train))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/loader/builder.py", line 49, in build_dataset
dataset = build(cfg_dataset, DATASETS, key="format")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/utils/build_utils.py", line 35, in build
return obj_cls(**cfg_copy)
^^^^^^^^^^^^^^^^^^^
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/loader/dataset/frame.py", line 59, in init
super().init(file_path, pipeline, data_prefix, test_mode)
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/loader/dataset/base.py", line 48, in init
self.info = self.load_file()
^^^^^^^^^^^^^^^^
File "/home/coderic/projects/action_paddle/paddle_video/PaddleVideo/paddlevideo/loader/dataset/frame.py", line 69, in load_file
frame_dir, frames_len, labels = line_split
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: too many values to unpack (expected 3)
LAUNCH INFO 2024-01-15 18:08:45,679 Exit code 1

How to replicate:

1. Dataset: https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/en/dataset/ucf101.md
2. Training: https://github.com/PaddlePaddle/PaddleVideo/blob/develop/docs/en/model_zoo/recognition/tsm.md

What seems to be the problem here?