Re-train using Resnet101

yijingru / Vertebra-Landmark-Detection

[ISBI 2020] Vertebra-Focused Landmark Detection for Scoliosis Assessment

MIT License

92 stars 26 forks source link

Re-train using Resnet101 #8

Open wahyurahmaniar opened 3 years ago

wahyurahmaniar commented 3 years ago

Hi, I want to re-train the images using ResNet101 But I got this error: RuntimeError: Given groups=1, weight of size [512, 1024, 3, 3], expected input[2, 2048, 64, 32] to have 1024 channels, but got 2048 channels instead

I just change this part in spinal_net.py: self.base_network = resnet.resnet101(pretrained=pretrained)

Do you have any suggestions? Thank you

yijingru commented 3 years ago

You may update resnet.py with the PytorchVersion you are using.

wahyurahmaniar commented 3 years ago

You may update resnet.py with the PytorchVersion you are using.

Hi, I change the PyTorch to 1.1.0, I still got the same error. Do you have tried to change the ResNet version?

Thank you