Change resnet type

[vigneshrk29]

Hello,

Thanks for your amazing work. In Posenet and Rootnet, if I change the resnet type value from 50 to 101 and 152, I get errors with the current snapshots. How do I work around this?

Errors are like some weights are missing

Thanks

[vigneshrk29]

For example,
Missing key(s) in state_dict: "module.backbone.layer3.6.conv1.weight", "module.backbone.layer3.6.bn1.weight", "module.backbone.layer3.6.bn1.bias", "module.backbone.layer3.6.bn1.running_mean", "module.backbone.layer3.6.bn1.running_var", "module.backbone.layer3.6.conv2.weight", "module.backbone.layer3.6.bn2.weight", "module.backbone.layer3.6.bn2.bias", "module.backbone.layer3.6.bn2.running_mean", "module.backbone.layer3.6.bn2.running_var", "module.backbone.layer3.6.conv3.weight", "module.backbone.layer3.6.bn3.weight", "module.backbone.layer3.6.bn3.bias", "module.backbone.layer3.6.bn3.running_mean", "module.backbone.layer3.6.bn3.running_var", "module.backbone.layer3.7.conv1.weight", "module.backbone.layer3.7.bn1.weight", "module.backbone.layer3.7.bn1.bias", "module.backbone.layer3.7.bn1.running_mean", "module.backbone.layer3.7.bn1.running_var", "module.backbone.layer3.7.conv2.weight", "module.backbone.layer3.7.bn2.weight", "module.backbone.layer3.7.bn2.bias", "module.backbone.layer3.7.bn2.running_mean", "module.backbone.layer3.7.bn2.running_var", "module.backbone.layer3.7.conv3.weight", "module.backbone.layer3.7.bn3.weight", "module.backbone.layer3.7.bn3.bias", "module.backbone.layer3.7.bn3.running_mean", "module.backbone.layer3.7.bn3.running_var", "module.backbone.layer3.8.conv1.weight", "module.backbone.layer3.8.bn1.weight", "module.backbone.layer3.8.bn1.bias", "module.backbone.layer3.8.bn1.running_mean", "module.backbone.layer3.8.bn1.running_var", "module.backbone.layer3.8.conv2.weight", "module.backbone.layer3.8.bn2.weight", "module.backbone.layer3.8.bn2.bias", "module.backbone.layer3.8.bn2.running_mean", "module.backbone.layer3.8.bn2.running_var", "module.backbone.layer3.8.conv3.weight", "module.backbone.layer3.8.bn3.weight", "module.backbone.layer3.8.bn3.bias", "module.backbone.layer3.8.bn3.running_mean", "module.backbone.layer3.8.bn3.running_var", "module.backbone.layer3.9.conv1.weight", "module.backbone.layer3.9.bn1.weight", "module.backbone.layer3.9.bn1.bias", "module.backbone.layer3.9.bn1.running_mean", "module.backbone.layer3.9.bn1.running_var", "module.backbone.layer3.9.conv2.weight", "module.backbone.layer3.9.bn2.weight", "module.backbone.layer3.9.bn2.bias", "module.backbone.layer3.9.bn2.running_mean", "module.backbone.layer3.9.bn2.running_var", "module.backbone.layer3.9.conv3.weight", "module.backbone.layer3.9.bn3.weight", "module.backbone.layer3.9.bn3.bias", "module.backbone.layer3.9.bn3.running_mean", "module.backbone.layer3.9.bn3.running_var", "module.backbone.layer3.10.conv1.weight", "module.backbone.layer3.10.bn1.weight", "module.backbone.layer3.10.bn1.bias", "module.backbone.layer3.10.bn1.running_mean", "module.backbone.layer3.10.bn1.running_var", "module.backbone.layer3.10.conv2.weight", "module.backbone.layer3.10.bn2.weight", "module.backbone.layer3.10.bn2.bias", "module.backbone.layer3.10.bn2.running_mean", "module.backbone.layer3.10.bn2.running_var", "module.backbone.layer3.10.conv3.weight", "module.backbone.layer3.10.bn3.weight", "module.backbone.layer3.10.bn3.bias", "module.backbone.layer3.10.bn3.running_mean", "module.backbone.layer3.10.bn3.running_var", "module.backbone.layer3.11.conv1.weight", "module.backbone.layer3.11.bn1.weight", "module.backbone.layer3.11.bn1.bias", "module.backbone.layer3.11.bn1.running_mean", "module.backbone.layer3.11.bn1.running_var", "module.backbone.layer3.11.conv2.weight", "module.backbone.layer3.11.bn2.weight", "module.backbone.layer3.11.bn2.bias", "module.backbone.layer3.11.bn2.running_mean", "module.backbone.layer3.11.bn2.running_var", "module.backbone.layer3.11.conv3.weight", "module.backbone.layer3.11.bn3.weight", "module.backbone.layer3.11.bn3.bias", "module.backbone.layer3.11.bn3.running_mean", "module.backbone.layer3.11.bn3.running_var", "module.backbone.layer3.12.conv1.weight", "module.backbone.layer3.12.bn1.weight", "module.backbone.layer3.12.bn1.bias", "module.backbone.layer3.12.bn1.running_mean", "module.backbone.layer3.12.bn1.running_var", "module.backbone.layer3.12.conv2.weight", "module.backbone.layer3.12.bn2.weight", "module.backbone.layer3.12.bn2.bias", "module.backbone.layer3.12.bn2.running_mean", "module.backbone.layer3.12.bn2.running_var", "module.backbone.layer3.12.conv3.weight", "module.backbone.layer3.12.bn3.weight", "module.backbone.layer3.12.bn3.bias", "module.backbone.layer3.12.bn3.running_mean", "module.backbone.layer3.12.bn3.running_var", "module.backbone.layer3.13.conv1.weight", "module.backbone.layer3.13.bn1.weight", "module.backbone.layer3.13.bn1.bias", "module.backbone.layer3.13.bn1.running_mean", "module.backbone.layer3.13.bn1.running_var", "module.backbone.layer3.13.conv2.weight", "module.backbone.layer3.13.bn2.weight", "module.backbone.layer3.13.bn2.bias", "module.backbone.layer3.13.bn2.running_mean", "module.backbone.layer3.13.bn2.running_var", "module.backbone.layer3.13.conv3.weight", "module.backbone.layer3.13.bn3.weight", "module.backbone.layer3.13.bn3.bias", "module.backbone.layer3.13.bn3.running_mean", "module.backbone.layer3.13.bn3.running_var", "module.backbone.layer3.14.conv1.weight", "module.backbone.layer3.14.bn1.weight", "module.backbone.layer3.14.bn1.bias", "module.backbone.layer3.14.bn1.running_mean", "module.backbone.layer3.14.bn1.running_var", "module.backbone.layer3.14.conv2.weight", "module.backbone.layer3.14.bn2.weight", "module.backbone.layer3.14.bn2.bias", "module.backbone.layer3.14.bn2.running_mean", "module.backbone.layer3.14.bn2.running_var", "module.backbone.layer3.14.conv3.weight", "module.backbone.layer3.14.bn3.weight", "module.backbone.layer3.14.bn3.bias", "module.backbone.layer3.14.bn3.running_mean", "module.backbone.layer3.14.bn3.running_var", "module.backbone.layer3.15.conv1.weight", "module.backbone.layer3.15.bn1.weight", "module.backbone.layer3.15.bn1.bias", "module.backbone.layer3.15.bn1.running_mean", "module.backbone.layer3.15.bn1.running_var", "module.backbone.layer3.15.conv2.weight", "module.backbone.layer3.15.bn2.weight", "module.backbone.layer3.15.bn2.bias", "module.backbone.layer3.15.bn2.running_mean", "module.backbone.layer3.15.bn2.running_var", "module.backbone.layer3.15.conv3.weight", "module.backbone.layer3.15.bn3.weight", "module.backbone.layer3.15.bn3.bias", "module.backbone.layer3.15.bn3.running_mean", "module.backbone.layer3.15.bn3.running_var", "module.backbone.layer3.16.conv1.weight", "module.backbone.layer3.16.bn1.weight", "module.backbone.layer3.16.bn1.bias", "module.backbone.layer3.16.bn1.running_mean", "module.backbone.layer3.16.bn1.running_var", "module.backbone.layer3.16.conv2.weight", "module.backbone.layer3.16.bn2.weight", "module.backbone.layer3.16.bn2.bias", "module.backbone.layer3.16.bn2.running_mean", "module.backbone.layer3.16.bn2.running_var", "module.backbone.layer3.16.conv3.weight", "module.backbone.layer3.16.bn3.weight", "module.backbone.layer3.16.bn3.bias", "module.backbone.layer3.16.bn3.running_mean", "module.backbone.layer3.16.bn3.running_var", "module.backbone.layer3.17.conv1.weight", "module.backbone.layer3.17.bn1.weight", "module.backbone.layer3.17.bn1.bias", "module.backbone.layer3.17.bn1.running_mean", "module.backbone.layer3.17.bn1.running_var", "module.backbone.layer3.17.conv2.weight", "module.backbone.layer3.17.bn2.weight", "module.backbone.layer3.17.bn2.bias", "module.backbone.layer3.17.bn2.running_mean", "module.backbone.layer3.17.bn2.running_var", "module.backbone.layer3.17.conv3.weight", "module.backbone.layer3.17.bn3.weight", "module.backbone.layer3.17.bn3.bias", "module.backbone.layer3.17.bn3.running_mean", "module.backbone.layer3.17.bn3.running_var", "module.backbone.layer3.18.conv1.weight", "module.backbone.layer3.18.bn1.weight", "module.backbone.layer3.18.bn1.bias", "module.backbone.layer3.18.bn1.running_mean", "module.backbone.layer3.18.bn1.running_var", "module.backbone.layer3.18.conv2.weight", "module.backbone.layer3.18.bn2.weight", "module.backbone.layer3.18.bn2.bias", "module.backbone.layer3.18.bn2.running_mean", "module.backbone.layer3.18.bn2.running_var", "module.backbone.layer3.18.conv3.weight", "module.backbone.layer3.18.bn3.weight", "module.backbone.layer3.18.bn3.bias", "module.backbone.layer3.18.bn3.running_mean", "module.backbone.layer3.18.bn3.running_var", "module.backbone.layer3.19.conv1.weight", "module.backbone.layer3.19.bn1.weight", "module.backbone.layer3.19.bn1.bias", "module.backbone.layer3.19.bn1.running_mean", "module.backbone.layer3.19.bn1.running_var", "module.backbone.layer3.19.conv2.weight", "module.backbone.layer3.19.bn2.weight", "module.backbone.layer3.19.bn2.bias", "module.backbone.layer3.19.bn2.running_mean", "module.backbone.layer3.19.bn2.running_var", "module.backbone.layer3.19.conv3.weight", "module.backbone.layer3.19.bn3.weight", "module.backbone.layer3.19.bn3.bias", "module.backbone.layer3.19.bn3.running_mean", "module.backbone.layer3.19.bn3.running_var", "module.backbone.layer3.20.conv1.weight", "module.backbone.layer3.20.bn1.weight", "module.backbone.layer3.20.bn1.bias", "module.backbone.layer3.20.bn1.running_mean", "module.backbone.layer3.20.bn1.running_var", "module.backbone.layer3.20.conv2.weight", "module.backbone.layer3.20.bn2.weight", "module.backbone.layer3.20.bn2.bias", "module.backbone.layer3.20.bn2.running_mean", "module.backbone.layer3.20.bn2.running_var", "module.backbone.layer3.20.conv3.weight", "module.backbone.layer3.20.bn3.weight", "module.backbone.layer3.20.bn3.bias", "module.backbone.layer3.20.bn3.running_mean", "module.backbone.layer3.20.bn3.running_var", "module.backbone.layer3.21.conv1.weight", "module.backbone.layer3.21.bn1.weight", "module.backbone.layer3.21.bn1.bias", "module.backbone.layer3.21.bn1.running_mean", "module.backbone.layer3.21.bn1.running_var", "module.backbone.layer3.21.conv2.weight", "module.backbone.layer3.21.bn2.weight", "module.backbone.layer3.21.bn2.bias", "module.backbone.layer3.21.bn2.running_mean", "module.backbone.layer3.21.bn2.running_var", "module.backbone.layer3.21.conv3.weight", "module.backbone.layer3.21.bn3.weight", "module.backbone.layer3.21.bn3.bias", "module.backbone.layer3.21.bn3.running_mean", "module.backbone.layer3.21.bn3.running_var", "module.backbone.layer3.22.conv1.weight", "module.backbone.layer3.22.bn1.weight", "module.backbone.layer3.22.bn1.bias", "module.backbone.layer3.22.bn1.running_mean", "module.backbone.layer3.22.bn1.running_var", "module.backbone.layer3.22.conv2.weight", "module.backbone.layer3.22.bn2.weight", "module.backbone.layer3.22.bn2.bias", "module.backbone.layer3.22.bn2.running_mean", "module.backbone.layer3.22.bn2.running_var", "module.backbone.layer3.22.conv3.weight", "module.backbone.layer3.22.bn3.weight", "module.backbone.layer3.22.bn3.bias", "module.backbone.layer3.22.bn3.running_mean", "module.backbone.layer3.22.bn3.running_var".

[mks0601]

Did you use --continue option? if so, you should not

[vigneshrk29]

Sorry, do you mean the continue train option?
I converted the demo.py into a jupyter notebook and ran it the same way as I did with resnet 50. For resnet 101, I just changed the config file (resnet-type)

[vigneshrk29]

Hi,

Any update on this?

[mks0601]

Are you trying to test with resnet101? If so, you should train a model with new resnet.