search

ArashHosseini / 3d-pose-baseline

A simple baseline for 3d human pose estimation in tensorflow. Presented at ICCV 17.
MIT License
302 stars 72 forks source link

Deal with case when input video contains numbers #2

Closed 09panesara closed 5 years ago

09panesara commented 5 years ago

re.findall followed by indexing by 0 was resulting in the wrong frame_indx for keypoint files obtained from videos with digits e.g. '001m_ang_2.6_win_1_000000000000_keypoints.json'.

Have modified to take the last match.