search

protossw512 / AdaptiveWingLoss

[ICCV 2019] Adaptive Wing Loss for Robust Face Alignment via Heatmap Regression - Official Implementation
Apache License 2.0
395 stars 88 forks source link

Usage of coordconv? #5

Closed PlumedSerpent closed 4 years ago

PlumedSerpent commented 4 years ago

Hi, in your model arch, you return outputs and boundary_channels with their sizes [B,landmarks+1,W,H] and [B,2,W,H]. I am not familiar with coordconv, should we calculate heatmap between concat([B,landmarks,W,H],[B,1,W,H]) and outputs[i]? If so, why should we return boundary_channels ? Thanks for your reply.

protossw512 commented 4 years ago

We do not need boundary_channels for inference, but we need for training since it is considered as one of the heatmap channels. A loss is applied to it against the groundtruth boundary channel.