Great work! I was trying to preprocess market1501 with https://github.com/Engineering-Course/LIP_JPPNet/, but I'm having a hard time getting your code to run... Are there any scripts you made to automate this? Thanks!

I am unable to train IAUNet for video Reid dataset. The repository does contain IAU3D architecture but it doesn't contain how to add IAU blocks in the resnet50 backbone for video Reid. Also will you please add complete code files for at least one video Reid dataset that I can train without any errors and confusion?

Also, I am having confusion that will part masks be also generated for videoreid dataset or not?

I need assistance in this regard

I want to insert IAU blocks into other pre-trained CNN architectures apart from ResNet50. How can I do that?

Need assistance in this regard

If I want to train the model on one dataset and test on another dataset, how can do that> I want to apply cross dataset validation. Need assistance in this regard

I want to train IAUnet for video reid datasets. I need assistance in this regard. Will you please share the complete github repo for the same?

SIAU

    x = z
    inputs = x
    b, c, h, w = x.size()
    u = x.view(b, c, -1).mean(2)  # b c 1

    a = self.sa(x) 
    x = torch.bmm(a.view(b, -1, h * w), x.view(b, c, -1).transpose(1, 2))   #b × C  × C
    x, u = self.reduce_dimension(x, u)
    y = self.SIAU(x, u) 

Spatial Semantic relation map may be HW × HW， however it is C × C as shown in code。