how tow concat roi feature with pose

garyzhao / SemGCN

The Pytorch implementation for "Semantic Graph Convolutional Networks for 3D Human Pose Regression" (CVPR 2019).

https://arxiv.org/abs/1904.03345

Apache License 2.0

461 stars 78 forks source link

how tow concat roi feature with pose #8

Open lingtengqiu opened 4 years ago

lingtengqiu commented 4 years ago

I am very interested in your work. In your paper, you said concate Roi feature with pose. Do you compress the feature to d-dimension? My understanding of your approach is： For example, pose(2D)+feature(d-dimen) == 2+D? Is it right? and what is your feature dimension. Looking forward to your reply.

garyzhao commented 4 years ago

Hi @lingtengqiu

Thanks for your interest in our work.

Yes, you are right. For example, we pool features from the conv_1 to conv_4 layers from ResNet. Therefore, let C_i be the channel size of the conv_i layer, and the pooled feature size will be (C_1 + C_2 + C_3 + C_4). After concatenated with the joint coordinate (x, y), the final feature size will be (C_1 + C_2 + C_3 + C_4 + 2). C_i may vary if you use different backbone (e.g., ResNet34 or ResNet101).

Best, Long