What is the usage of edge detection in 'HardDetectionModule' in model_test.py?

[chaffeechenyefei]

Thanks for sharing such good implementation!
After learning from the code, I was confused about the edge detection part in 'HardDetectionModule' in file 'model_test.py'. The code goes like:
detected = torch.min( is_depth_wise_max, torch.min(is_local_max, is_not_edge) )
Why edges should not be consider as detection? Why feature extracted by VGG16 can be used as the input of edge detection? What's the idea behind it?

[chaffeechenyefei]

Oh, sorry for that. I have found the answer in SIFT paper: 'Eliminating edge responses'. Thanks for sharing code again.

[mihaidusmanu]

Hello, yes exactly - as for SIFT, we have eliminated the edge responses since they are generally less stable. However, in practice for D2-Net features, this check doesn't make a big difference: if I recall correctly, most of the times ~200 keypoints are discarded because of the edge filtering.

[chaffeechenyefei]

Thanks for the explanation. Really appreciate that.