a problem of training model

[chen-ming2019]

I met a problem named as "FileNotFoundError: [Errno 2] No such file or directory: 'dataset\UCF-101\WalkingWithDog\img_00001.jpg' when I trained the model using "run_ECOLite_finetune_UCF101.sh".It couldn't continue to run with "train(train_loader, model, criterion, optimizer, epoch)" using debug.I could't find the reason resulting in this. can you help me ? Thanks!

[LuyuanLi97]

Hi! Have you solved this problem? I also met a problem as "FileNotFoundError: [Errno 2] No such file or directory: '../UCF101_frames/v_BlowingCandles_g21_c04/img_00001.jpg'" when running main.py.

Update: I solved this problem. Make sure that you use absolute path instead of relative path. Also, remember to change the "--rgb_prefix" argument into the prefix of your dataset name.

[chen-ming2019]

@LuyuanLi97 Hi! The path and '--rgb_prefix' what you said is in the main.py or gen_dataset_list.py.In addition,are there only videos in your ucf101 dataset?

[LuyuanLi97]

@chen-ming2019 '--rgb_prefix' is the argument in main.py. The dataset is not videos but are organized as dataset_root_dir/video_names/frame_pictures, if the frame pictures in your dataset has name "image_XXX", than you should run like ''--rgb_prefix image_''

[chen-ming2019]

@LuyuanLi97 How can you obtain the frame pictures from original video(dataset)?Seperate the video into frames（frame by frame）?

[LuyuanLi97]

@chen-ming2019 Maybe you can refer to this script: https://github.com/FingerRec/activity_recognition/blob/a899b57801/data/video_jpg_ucf101_hmdb51.py

[chen-ming2019]

@LuyuanLi97 ok ! Thank you very much!

[chen-ming2019]

@LuyuanLi97 Hi Luyuan,Have you done the expermient in the something-something-v1? When I downloaded the dataset and unziped it using the order"cat ...| tar zx",some errors emerged called "gzip: stdin: unexpected end of file ;tar : 归档文件中的异常 EOF".After that,I download the dataset again ,but unfortunnately the problem has been existing.