The difference between PEG and PEG for detection

[mt-cly]

Hi, thanks for your great work. I meet some confusion when reading the paper. Specifically, in the part of Supplement C. Example Code, there is a light difference between the two presented Algorithms. In Algorithm 1 PyTorch snippet of PEG, the PEG includes a Conv layer, while in Algorithm 2PyTorch snippet of PEG for detection, there are additional BN+Relu layers. I wonder how about the effectiveness comparison of this two setting, would the second setting with BN+Relu be better? Thank you.