search

sunnyxiaohu / R-C3D.pytorch

R-C3D pytorch implementation
248 stars 67 forks source link

Inference the model use my own video #37

Closed dagongji10 closed 4 years ago

dagongji10 commented 5 years ago

@sunnyxiaohu I want to use R-C3D.pytorch to get the performance. The dataset ThUMOS14 is too large, so I want to just use less untrimmed-video defined by myself to test the model. Is it realistic? How can I realize it?

Maruidear commented 4 years ago

Have you solved the problem?

dagongji10 commented 4 years ago

Have you solved the problem?

No， I use TSN instead of R-C3D to do action_recognize.

Maruidear commented 4 years ago

你解决了这个问题吗？

不，我用TSN代替R-C3D进行action_recognize。

Have you successfully converted your own data set with TSN?

dagongji10 commented 4 years ago

你解决了这个问题吗？

不，我用TSN代替R-C3D进行action_recognize。

Have you successfully converted your own data set with TSN?

Yeah, I use TSN to extract 1024-dim action feature, it seems work.

Maruidear commented 4 years ago

你解决了这个问题吗？

不，我用TSN代替R-C3D进行action_recognize。

您是否已使用TSN成功转换了自己的数据集？

是的，我使用TSN提取1024-dim动作功能，看来可行。

Thank you very much for your reply.