Question about GAST-Net model

fabro66 / GAST-Net-3DPoseEstimation

A Graph Attention Spatio-temporal Convolutional Networks for 3D Human Pose Estimation in Video (GAST-Net)

MIT License

312 stars 70 forks source link

Question about GAST-Net model #44

Open z0978916348 opened 3 years ago

z0978916348 commented 3 years ago

Hi! Can you explain why GAST-Net's dimension is (25, 17, 256) (T, N, C) after executing first Graph Attention Block in the paper? I think input shape (256, 27, 17) will be (256, 9, 17) after going through Conv2D with kernel (3,1) and stride (3, 1) in gast_net.py When I print the residual shape in the network, it shows me (256, 9, 17) # (C, T, N).

Thanks