QizeYang/PAUL

pre-train model

Opened this issue · 1 comments

what data set did you train the model on the driver? it is MSMT17, is this right?. I want to use this model but there is an error '../../imagenet_models/resnet50-19c8e357.pth'.
thank you!

Hi! You are right and that error was caused by the default value of the parameters of PatchNetUn. If possible, you can change pretrained=True to pretrained=False.

def __init__(self, class_num=1000, backbone=resnet50, pretrained=False, is_for_test=False,

Thank you.