Merge pull request #2 from insikk/resnet

Resnet

.gitignore

@@ -0,0 +1,8 @@
+# model weight
+*.ckpt
+
+# Python junk
+__pycache__
+.ipynb_checkpoints
+*.npy
+*.pyc

README.md

@@ -1,12 +1,18 @@
 # Grad-CAM-tensorflow
 
-This is tensorflow version of demo for Grad-CAM. I used vgg16 for demo because this model is very popular CNN model.
-However grad-cam can be used with any CNN model. Just modify convolution layer in my demo code.
+This is tensorflow version of demo for Grad-CAM. I used ResNet-v1-101, ResNet-v1-50, and vgg16 for demo because this models are very popular CNN model.
+However grad-cam can be used with any other CNN models. Just modify convolution layer in my demo code.
 
 ![Preview](https://github.com/insikk/Grad-CAM-tensorflow/blob/master/image_preview.png?raw=true)
 
 See [python notebook](https://github.com/insikk/Grad-CAM-tensorflow/blob/master/gradCAM_tensorflow_demo.ipynb) to see demo of this repository.
->To use the VGG networks in this demo, the npy files for [VGG16 NPY](https://mega.nz/#!YU1FWJrA!O1ywiCS2IiOlUCtCpI6HTJOMrneN-Qdv3ywQP5poecM) has to be downloaded.
+>To use VGG networks in this demo, the npy files for [VGG16 NPY](ftp://mi.eng.cam.ac.uk/pub/mttt2/models/vgg16.npy) has to be downloaded.
+
+>To use ResNet-v1-50 or ResNet-v1-101, download weight from https://github.com/tensorflow/models/tree/master/research/slim
+
+
+**Any Contributions are Welcome**
+
 
 ## [Origial Paper] Grad-CAM: Gradient-weighted Class Activation Mapping
 
@@ -16,6 +22,15 @@ Ramprasaath R. Selvaraju, Abhishek Das, Ramakrishna Vedantam, Michael Cogswell,
 
 ![Overview](http://i.imgur.com/JaGbdZ5.png)
 
+# Requirements
+
+* GPU Memory: 6GB or higher to run VGG16, and ResNet101 (You may able to run ResNet50 with less than 6GB)
+
+# Setup
+
+```
+export PYTHONPATH=$PYTHONPATH:`pwd`/slim
+```
 
 ## Acknowledgement
 

model/vgg16.py

@@ -17,7 +17,7 @@ def __init__(self, vgg16_npy_path=None, trainable=True):
             vgg16_npy_path = path
             print(path)
 
-        self.data_dict = np.load(vgg16_npy_path, encoding='latin1').item()        
+        self.data_dict = np.load(vgg16_npy_path, encoding='latin1').item()
         self.trainable = trainable
         print("npy file loaded")
 
@@ -33,15 +33,15 @@ def build(self, rgb, train_mode=None):
         rgb_scaled = rgb * 255.0
 
         # Convert RGB to BGR
-        red, green, blue = tf.split(3, 3, rgb_scaled)
+        red, green, blue = tf.split(rgb_scaled, 3, 3)
         assert red.get_shape().as_list()[1:] == [224, 224, 1]
         assert green.get_shape().as_list()[1:] == [224, 224, 1]
         assert blue.get_shape().as_list()[1:] == [224, 224, 1]
-        bgr = tf.concat(3, [
+        bgr = tf.concat([
             blue - VGG_MEAN[0],
             green - VGG_MEAN[1],
             red - VGG_MEAN[2],
-        ])
+        ], 3)
         assert bgr.get_shape().as_list()[1:] == [224, 224, 3]
 
         self.conv1_1 = self.conv_layer(bgr, "conv1_1")
@@ -70,7 +70,7 @@ def build(self, rgb, train_mode=None):
         self.fc6 = self.fc_layer(self.pool5, "fc6")
         assert self.fc6.get_shape().as_list()[1:] == [4096]
         self.relu6 = tf.nn.relu(self.fc6)
-        
+
         if train_mode is not None:
             self.relu6 = tf.cond(train_mode, lambda: tf.nn.dropout(self.relu6, 0.5), lambda: self.relu6)
         elif self.trainable:
@@ -79,7 +79,7 @@ def build(self, rgb, train_mode=None):
 
         self.fc7 = self.fc_layer(self.relu6, "fc7")
         self.relu7 = tf.nn.relu(self.fc7)
-        
+
         if train_mode is not None:
             self.relu7 = tf.cond(train_mode, lambda: tf.nn.dropout(self.relu7, 0.5), lambda: self.relu7)
         elif self.trainable:
@@ -134,4 +134,5 @@ def get_bias(self, name):
         return tf.Variable(self.data_dict[name][1], name="biases")
 
     def get_fc_weight(self, name):
-        return tf.Variable(self.data_dict[name][0], name="weights")
+        return tf.Variable(self.data_dict[name][0], name="weights")
+