W-Net is almost the same as paper "Dual Path Multi-Scale Fusion Networks with Attention for Crowd Counting"

[pikerbright]

I found that w-net network is actually the same as SFANet in paper "Dual Path Multi-Scale Fusion Networks with Attention for Crowd Counting". Some hyper parameters and training details are also the same.

Can someone help to review these papers?

[hellowangqian]

Creepy, based on the arXiv-ed time stamps, SFANet seems to have appeared earlier.