I want to try this code on some videos, but the results of using demo multi-person is not good, the joints fluctuate a lot. Also, there is no pose_seq_net.py in nnet folder when I set the parameter, cfg.video = True.
May I ask is it possible to publish some examples to process videos, or any suggestion I can implement it myself?
Hello,
Thanks a lot for this great work.
I want to try this code on some videos, but the results of using demo multi-person is not good, the joints fluctuate a lot. Also, there is no pose_seq_net.py in nnet folder when I set the parameter, cfg.video = True.
May I ask is it possible to publish some examples to process videos, or any suggestion I can implement it myself?
Thanks again.