Question on evaluation

[andyj1]

def load_pretrained_weights(self, path): w_dict = torch.load(path + "/backbone", map_location=lambda storage, loc: storage) self.backbone.load_state_dict(w_dict, strict=True) w_dict = torch.load(path + "/segmentation_decoder", map_location=lambda storage, loc: storage) self.decoder.load_state_dict(w_dict, strict=True)

thanks for the amazing work. I have a question on running the evaluation code. The code above in the modules/rangenet/model.py (for RangeNet++ for example), it directs to './pretrained_weights/kitti/rangenet/backbone', which is a directory. I wonder how to properly load the pretrained_weights. Thanks!

[yiningsutd]

Met the same prob. Replacing rangenet folder with ckpt from this url fixs this prob.

[cqf7419]

def load_pretrained_weights(self, path): w_dict = torch.load(path + "/backbone", map_location=lambda storage, loc: storage) self.backbone.load_state_dict(w_dict, strict=True) w_dict = torch.load(path + "/segmentation_decoder", map_location=lambda storage, loc: storage) self.decoder.load_state_dict(w_dict, strict=True)

thanks for the amazing work. I have a question on running the evaluation code. The code above in the modules/rangenet/model.py (for RangeNet++ for example), it directs to './pretrained_weights/kitti/rangenet/backbone', which is a directory. I wonder how to properly load the pretrained_weights. Thanks!

w_dict = torch.load(path + "/backbone.zip" , map_location=lambda storage, loc: storage)

I make it by add '.zip' suffix.