ResNet-152 added

deeplabcut/pose_cfg.yaml

@@ -47,7 +47,7 @@ bottomheight: 400
 
 
 
-# Type of the CNN to use, currently resnet_101 and resnet_50
+# Type of the CNN to use, currently resnet_152, resnet_101 and resnet_50
 # are supported
 net_type: resnet_50
 

deeplabcut/pose_estimation_tensorflow/models/pretrained/download.sh

@@ -3,3 +3,4 @@
 
 curl http://download.tensorflow.org/models/resnet_v1_50_2016_08_28.tar.gz | tar xvz
 curl http://download.tensorflow.org/models/resnet_v1_101_2016_08_28.tar.gz | tar xvz
+curl http://download.tensorflow.org/models/resnet_v1_152_2016_08_28.tar.gz | tar xvz

deeplabcut/pose_estimation_tensorflow/models/pretrained/pretrained_model_urls.yaml

@@ -1,2 +1,3 @@
 resnet_50: http://download.tensorflow.org/models/resnet_v1_50_2016_08_28.tar.gz
 resnet_101: http://download.tensorflow.org/models/resnet_v1_101_2016_08_28.tar.gz
+resnet_152: http://download.tensorflow.org/models/resnet_v1_152_2016_08_28.tar.gz

deeplabcut/pose_estimation_tensorflow/nnet/pose_net.py

@@ -12,7 +12,8 @@
 
 
 net_funcs = {'resnet_50': resnet_v1.resnet_v1_50,
-             'resnet_101': resnet_v1.resnet_v1_101}
+             'resnet_101': resnet_v1.resnet_v1_101,
+             'resnet_152': resnet_v1.resnet_v1_152}
 
 
 def prediction_layer(cfg, input, name, num_outputs):

deeplabcut/utils/auxfun_models.py

@@ -14,8 +14,10 @@ def Check4weights(modeltype,parent_path,num_shuffles):
         model_path = parent_path  / 'pose_estimation_tensorflow/models/pretrained/resnet_v1_50.ckpt'
     elif 'resnet_101' == modeltype:
         model_path = parent_path / 'pose_estimation_tensorflow/models/pretrained/resnet_v1_101.ckpt'
+    elif 'resnet_152' == modeltype:
+        model_path = parent_path / 'pose_estimation_tensorflow/models/pretrained/resnet_v1_152.ckpt'
     else:
-        print("Currently only ResNet 50 or 101 supported, please change 'resnet' entry in config.yaml!")
+        print("Currently only ResNet 50, 101 or 152 supported, please change 'resnet' entry in config.yaml!")
         num_shuffles=-1 #thus the loop below is empty...
         model_path=parent_path
 

docs/functionDetails.md

@@ -156,7 +156,7 @@ Specifically, the user can edit the **pose_cfg.yaml** within the **train** subdi
 configuration files contain meta information with regard to the parameters of the feature detectors. Key parameters
 are listed in Box 2.
 
-At this step, the ImageNet pre-trained ResNet-50 and ResNet-101 weights will be downloaded. If they do not download (you will see this downloading in the terminal, then you may not have permission to do so (something we have seen with some Windows users - see the **[WIKI troubleshooting for more help!](https://github.com/AlexEMG/DeepLabCut/wiki/Troubleshooting-Tips)**). If you are labeling adult human data, you may also want to use a human-pretrained network. A ResNet-101 pre-trained on MPII is available. Download it and move to the pretrained folder (this will be in your site-packages, under ``../pose-tensorflow/models/pretrained``). Download with the following commands in the terminal: 
+At this step, the ImageNet pre-trained ResNet-50, ResNet-101 and ResNet-152 weights will be downloaded. If they do not download (you will see this downloading in the terminal, then you may not have permission to do so (something we have seen with some Windows users - see the **[WIKI troubleshooting for more help!](https://github.com/AlexEMG/DeepLabCut/wiki/Troubleshooting-Tips)**). If you are labeling adult human data, you may also want to use a human-pretrained network. A ResNet-101 pre-trained on MPII is available. Download it and move to the pretrained folder (this will be in your site-packages, under ``../pose-tensorflow/models/pretrained``). Download with the following commands in the terminal: 
  ```
     curl -L -O https://datasets.d2.mpi-inf.mpg.de/deepercut-models-tensorflow/mpii-single-resnet-101.data-00000-of-00001 
     curl -L -O https://datasets.d2.mpi-inf.mpg.de/deepercut-models-tensorflow/mpii-single-resnet-101.meta