search

agethen / RPAN

Our implementation of Recurrent Pose Attention in Du et al.: "RPAN: An End-to-End Recurrent Pose-attention Network for Action Recognition in Videos"
37 stars 15 forks source link

about the implementation result #2

Closed demianzhang closed 5 years ago

demianzhang commented 5 years ago

Could you tell me the finally result comparing with the original paper. Many thanks.

agethen commented 5 years ago

Hi,

I did run it on JHMDB (RGB, split 1), where I got 61.3% Top-1 accuracy. The paper got 60.0, but I think the authors did not use ResNet as a base. For comparison, I got 59.3% for a ResNet + LSTM + Classifier baseline.

demianzhang commented 5 years ago

Thanks for your reply.