search

ZhaofanQiu / pseudo-3d-residual-networks

Pseudo-3D Convolutional Residual Networks for Video Representation Learning
MIT License
352 stars 119 forks source link

Question about classifier #29

Open bbp94 opened 5 years ago

bbp94 commented 5 years ago

Thank you for your grate work! I have some questions about classifier：Whether the experiments in your paper were implemented by P3D+SVM？If so，did you use softmax to optimize the P3D until convergence and then get the final feature to train SVM for final predict result？

bbp94 commented 5 years ago

I used softmax as classifier and got top1 accuracy 81%，is this correct？