search

SydCaption / SAAT

MIT License
62 stars 21 forks source link

Is it right to use 3D-ResNets-Pytorch for 3D motion features extraction? #35

Closed MarcusNerva closed 3 years ago

MarcusNerva commented 3 years ago

Hello amigo! I have a question about feature extraction for 3D motion features. In the paper Syntax-Aware Action Targeting for Video Captioning, the author declares that he used C3D for motion features extraction. However, I am a little bit confused about the difference between '3D-ResNets' and 'C3D'. Are these two the same thing? Looking forward to your reply!

SydCaption commented 3 years ago

Hi, basically, you can use whatever conv3D models to extract the motion features. I did experiments on 3D-ResNets, and I didn't try other 3D feature extractors.

MarcusNerva commented 3 years ago

Hi, basically, you can use whatever conv3D models to extract the motion features. I did experiments on 3D-ResNets, and I didn't try other 3D feature extractors.

Thank you for your quick reply. But you misunderstand what I mean. I was wondering whether '3D-ResNets' and 'C3D' are the same thing?

SydCaption commented 3 years ago

Hi, I think C3D means 3D-CNN, so 3D-ResNets is a type of C3D.

MarcusNerva commented 3 years ago

Hi, I think C3D means 3D-CNN, so 3D-ResNets is a type of C3D.

How about I3D? It seems that I3D is also a type of 3D-CNN.

SydCaption commented 3 years ago

Exactly!

MarcusNerva @.***>于2021年7月14日 周三下午6:51写道：

Hi, I think C3D means 3D-CNN, so 3D-ResNets is a type of C3D.

How about I3D? It seems that I3D is also a type of 3D-CNN.

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub https://github.com/SydCaption/SAAT/issues/35#issuecomment-879716609, or unsubscribe https://github.com/notifications/unsubscribe-auth/AO3DDOIQVSFU73YMS7L7CFTTXVFYHANCNFSM5AK4RT5A .