search

yjxiong / temporal-segment-networks

Code & Models for Temporal Segment Networks (TSN) in ECCV 2016
BSD 2-Clause "Simplified" License
1.53k stars 477 forks source link

what's the UCF101 structure when extract optical flow? #262

Closed Ixiaohuihuihui closed 4 years ago

Ixiaohuihuihui commented 5 years ago

When I run the command "bash scripts/extract_optical_flow.sh", I got the output message: Extracting optical flow from videos in folder: ../Dataset/UCF-101 reading videos from folder: ../Dataset/UCF-101 selected extension of videos: avi total number of videos found: 0

can you tell me what's the UCF-101 dataset? I mean the original dataset structure is UCF101 / ApplyEyeMakeup/ v_ApplyEyeMakeup_g01_c01.avi In the file 'build_of.py', vid_list = glob.glob(src_path+'/*.'+ext) so I need to change the dataset structure to two level structure？ UCF101 / v_ApplyEyeMakeup_g01_c01.avi

Ixiaohuihuihui commented 5 years ago

Thanks very much! I have solved this problem. I changed the line 120 of 'build_of.py'. vid_list = glob.glob(src_path+'/*/*.'+ext)