search

machanic / AU_R-CNN

The official implementation code of paper: "AU R-CNN：Encoding Expert Prior Knowledge into R-CNN for Action Unit Detection".
https://arxiv.org/abs/1812.05788
87 stars 24 forks source link

Where is "ResNet-101-model.npz" ? #2

Closed angleboy8 closed 3 years ago

angleboy8 commented 4 years ago

I tested your demo, but I didn't find the ResNet-101-model.npz. I hope you can provide this model. Thank you!

machanic commented 4 years ago

The ResNet-101-model.npz is loading the ImageNet pretrained file at the first training. It is converted from caffe model file. You can check this page and source code to see how to download and convert. https://docs.chainer.org/en/stable/reference/generated/chainer.links.ResNet101Layers.html#chainer.links.ResNet101Layers Also I will upload the npz file into Google Drive so that you can directly download.

angleboy8 commented 4 years ago

I appreciate your help. I hope you will upload your google cloud earlier.

machanic commented 4 years ago

@angleboy8 I have uploaded all the training necessary files onto https://github.com/sharpstill/AU_R-CNN/tree/master/necessary_train_files_for_AU_R-CNN

rswethashankar92 commented 4 years ago

while testing the demo, i think i need to change the weight_params in HeatMapGenerator class, please help me

machanic commented 4 years ago

@rswethashankar92 Which .py file do you use? Can you describe more detail?

machanic commented 3 years ago

@angleboy8 All the pretrained model file(.npz) are located here : https://cloud.tsinghua.edu.cn/d/8190d64e83ef43b7ab3f/?p=%2FBP4D_dataset_trained_model&mode=list