Improved-re-identification

I am making modifications to the re-identification model described in the paper An Improved Deep Learning Architecture for Person Re-Identification.

run.py is the original file without attention.

In run2.py I have implemented visual attention. In run3.py I am currently in the process of implementing channel attention inspired by the paper ABD-Net: Attentive but Diverse Person Re-Identification

Here is my design for the visual attention part.