You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2018/11/15 23:05:38 UTC
[GitHub] indhub closed pull request #13255: [Example] Gradcam consolidation
in tutorial
indhub closed pull request #13255: [Example] Gradcam consolidation in tutorial
URL: https://github.com/apache/incubator-mxnet/pull/13255
This is a PR merged from a forked repository.
As GitHub hides the original diff on merge, it is displayed below for
the sake of provenance:
As this is a foreign pull request (from a fork), the diff is supplied
below (as it won't show otherwise due to GitHub magic):
diff --git a/docs/conf.py b/docs/conf.py
index 656a1da96d6..af235219e4f 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -107,7 +107,7 @@
# List of patterns, relative to source directory, that match files and
# directories to ignore when looking for source files.
-exclude_patterns = ['3rdparty', 'build_version_doc', 'virtualenv', 'api/python/model.md', 'README.md']
+exclude_patterns = ['3rdparty', 'build_version_doc', 'virtualenv', 'api/python/model.md', 'README.md', 'tutorial_utils']
# The reST default role (used for this markup: `text`) to use for all documents.
#default_role = None
diff --git a/example/cnn_visualization/gradcam.py b/docs/tutorial_utils/vision/cnn_visualization/gradcam.py
similarity index 100%
rename from example/cnn_visualization/gradcam.py
rename to docs/tutorial_utils/vision/cnn_visualization/gradcam.py
diff --git a/docs/tutorials/vision/cnn_visualization.md b/docs/tutorials/vision/cnn_visualization.md
index a350fffaa36..fd6a464e99d 100644
--- a/docs/tutorials/vision/cnn_visualization.md
+++ b/docs/tutorials/vision/cnn_visualization.md
@@ -22,7 +22,7 @@ from matplotlib import pyplot as plt
import numpy as np
gradcam_file = "gradcam.py"
-base_url = "https://raw.githubusercontent.com/indhub/mxnet/cnnviz/example/cnn_visualization/{}?raw=true"
+base_url = "https://github.com/apache/incubator-mxnet/tree/master/docs/tutorial_utils/vision/cnn_visualization/{}?raw=true"
mx.test_utils.download(base_url.format(gradcam_file), fname=gradcam_file)
import gradcam
```
@@ -182,6 +182,7 @@ Next, we'll write a method to get an image, preprocess it, predict category and
2. **Guided Grad-CAM:** Guided Grad-CAM shows which exact pixels contributed the most to the CNN's decision.
3. **Saliency map:** Saliency map is a monochrome image showing which pixels contributed the most to the CNN's decision. Sometimes, it is easier to see the areas in the image that most influence the output in a monochrome image than in a color image.
+
```python
def visualize(net, img_path, conv_layer_name):
orig_img = mx.img.imread(img_path)
diff --git a/example/cnn_visualization/README.md b/example/cnn_visualization/README.md
deleted file mode 100644
index 10b91492600..00000000000
--- a/example/cnn_visualization/README.md
+++ /dev/null
@@ -1,17 +0,0 @@
-# Visualzing CNN decisions
-
-This folder contains an MXNet Gluon implementation of [Grad-CAM](https://arxiv.org/abs/1610.02391) that helps visualize CNN decisions.
-
-A tutorial on how to use this from Jupyter notebook is available [here](https://mxnet.incubator.apache.org/tutorials/vision/cnn_visualization.html).
-
-You can also do the visualization from terminal:
-```
-$ python gradcam_demo.py hummingbird.jpg
-Predicted category : hummingbird (94)
-Original Image : hummingbird_orig.jpg
-Grad-CAM : hummingbird_gradcam.jpg
-Guided Grad-CAM : hummingbird_guided_gradcam.jpg
-Saliency Map : hummingbird_saliency.jpg
-```
-
-
diff --git a/example/cnn_visualization/gradcam_demo.py b/example/cnn_visualization/gradcam_demo.py
deleted file mode 100644
index d9ca5ddade8..00000000000
--- a/example/cnn_visualization/gradcam_demo.py
+++ /dev/null
@@ -1,110 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements. See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership. The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License. You may obtain a copy of the License at
-#
-# http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied. See the License for the
-# specific language governing permissions and limitations
-# under the License.
-
-import mxnet as mx
-from mxnet import gluon
-
-import argparse
-import os
-import numpy as np
-import cv2
-
-import vgg
-import gradcam
-
-# Receive image path from command line
-parser = argparse.ArgumentParser(description='Grad-CAM demo')
-parser.add_argument('img_path', metavar='image_path', type=str, help='path to the image file')
-
-args = parser.parse_args()
-
-# We'll use VGG-16 for visualization
-network = vgg.vgg16(pretrained=True, ctx=mx.cpu())
-# We'll resize images to 224x244 as part of preprocessing
-image_sz = (224, 224)
-
-def preprocess(data):
- """Preprocess the image before running it through the network"""
- data = mx.image.imresize(data, image_sz[0], image_sz[1])
- data = data.astype(np.float32)
- data = data/255
- # These mean values were obtained from
- # https://mxnet.incubator.apache.org/api/python/gluon/model_zoo.html
- data = mx.image.color_normalize(data,
- mean=mx.nd.array([0.485, 0.456, 0.406]),
- std=mx.nd.array([0.229, 0.224, 0.225]))
- data = mx.nd.transpose(data, (2,0,1)) # Channel first
- return data
-
-def read_image_mxnet(path):
- with open(path, 'rb') as fp:
- img_bytes = fp.read()
- return mx.img.imdecode(img_bytes)
-
-def read_image_cv(path):
- return cv2.resize(cv2.cvtColor(cv2.imread(path), cv2.COLOR_BGR2RGB), image_sz)
-
-# synset.txt contains the names of Imagenet categories
-# Load the file to memory and create a helper method to query category_index -> category name
-synset_url = "http://data.mxnet.io/models/imagenet/synset.txt"
-synset_file_name = "synset.txt"
-mx.test_utils.download(synset_url, fname=synset_file_name)
-
-synset = []
-with open('synset.txt', 'r') as f:
- synset = [l.rstrip().split(' ', 1)[1].split(',')[0] for l in f]
-
-def get_class_name(cls_id):
- return "%s (%d)" % (synset[cls_id], cls_id)
-
-def run_inference(net, data):
- """Run the input image through the network and return the predicted category as integer"""
- out = net(data)
- return out.argmax(axis=1).asnumpy()[0].astype(int)
-
-def visualize(net, img_path, conv_layer_name):
- """Create Grad-CAM visualizations using the network 'net' and the image at 'img_path'
- conv_layer_name is the name of the top most layer of the feature extractor"""
- image = read_image_mxnet(img_path)
- image = preprocess(image)
- image = image.expand_dims(axis=0)
-
- pred_str = get_class_name(run_inference(net, image))
-
- orig_img = read_image_cv(img_path)
- vizs = gradcam.visualize(net, image, orig_img, conv_layer_name)
- return (pred_str, (orig_img, *vizs))
-
-# Create Grad-CAM visualization for the user provided image
-last_conv_layer_name = 'vgg0_conv2d12'
-cat, vizs = visualize(network, args.img_path, last_conv_layer_name)
-
-print("{0:20}: {1:80}".format("Predicted category", cat))
-
-# Write the visualiations into file
-img_name = os.path.split(args.img_path)[1].split('.')[0]
-suffixes = ['orig', 'gradcam', 'guided_gradcam', 'saliency']
-image_desc = ['Original Image', 'Grad-CAM', 'Guided Grad-CAM', 'Saliency Map']
-
-for i, img in enumerate(vizs):
- img = img.astype(np.float32)
- if len(img.shape) == 3:
- img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
- out_file_name = "%s_%s.jpg" % (img_name, suffixes[i])
- cv2.imwrite(out_file_name, img)
- print("{0:20}: {1:80}".format(image_desc[i], out_file_name))
-
diff --git a/example/cnn_visualization/vgg.py b/example/cnn_visualization/vgg.py
deleted file mode 100644
index a8a0ef6c8de..00000000000
--- a/example/cnn_visualization/vgg.py
+++ /dev/null
@@ -1,90 +0,0 @@
-# Licensed to the Apache Software Foundation (ASF) under one
-# or more contributor license agreements. See the NOTICE file
-# distributed with this work for additional information
-# regarding copyright ownership. The ASF licenses this file
-# to you under the Apache License, Version 2.0 (the
-# "License"); you may not use this file except in compliance
-# with the License. You may obtain a copy of the License at
-#
-# http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing,
-# software distributed under the License is distributed on an
-# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-# KIND, either express or implied. See the License for the
-# specific language governing permissions and limitations
-# under the License.
-
-import mxnet as mx
-from mxnet import gluon
-
-import os
-from mxnet.gluon.model_zoo import model_store
-
-from mxnet.initializer import Xavier
-from mxnet.gluon.nn import MaxPool2D, Flatten, Dense, Dropout, BatchNorm
-from gradcam import Activation, Conv2D
-
-class VGG(mx.gluon.HybridBlock):
- def __init__(self, layers, filters, classes=1000, batch_norm=False, **kwargs):
- super(VGG, self).__init__(**kwargs)
- assert len(layers) == len(filters)
- with self.name_scope():
- self.features = self._make_features(layers, filters, batch_norm)
- self.features.add(Dense(4096, activation='relu',
- weight_initializer='normal',
- bias_initializer='zeros'))
- self.features.add(Dropout(rate=0.5))
- self.features.add(Dense(4096, activation='relu',
- weight_initializer='normal',
- bias_initializer='zeros'))
- self.features.add(Dropout(rate=0.5))
- self.output = Dense(classes,
- weight_initializer='normal',
- bias_initializer='zeros')
-
- def _make_features(self, layers, filters, batch_norm):
- featurizer = mx.gluon.nn.HybridSequential(prefix='')
- for i, num in enumerate(layers):
- for _ in range(num):
- featurizer.add(Conv2D(filters[i], kernel_size=3, padding=1,
- weight_initializer=Xavier(rnd_type='gaussian',
- factor_type='out',
- magnitude=2),
- bias_initializer='zeros'))
- if batch_norm:
- featurizer.add(BatchNorm())
- featurizer.add(Activation('relu'))
- featurizer.add(MaxPool2D(strides=2))
- return featurizer
-
- def hybrid_forward(self, F, x):
- x = self.features(x)
- x = self.output(x)
- return x
-
-vgg_spec = {11: ([1, 1, 2, 2, 2], [64, 128, 256, 512, 512]),
- 13: ([2, 2, 2, 2, 2], [64, 128, 256, 512, 512]),
- 16: ([2, 2, 3, 3, 3], [64, 128, 256, 512, 512]),
- 19: ([2, 2, 4, 4, 4], [64, 128, 256, 512, 512])}
-
-def get_vgg(num_layers, pretrained=False, ctx=mx.cpu(),
- root=os.path.join('~', '.mxnet', 'models'), **kwargs):
- layers, filters = vgg_spec[num_layers]
- net = VGG(layers, filters, **kwargs)
- net.initialize(ctx=ctx)
-
- # Get the pretrained model
- vgg = mx.gluon.model_zoo.vision.get_vgg(num_layers, pretrained=True, ctx=ctx)
-
- # Set the parameters in the new network
- params = vgg.collect_params()
- for key in params:
- param = params[key]
- net.collect_params()[net.prefix+key.replace(vgg.prefix, '')].set_data(param.data())
-
- return net
-
-def vgg16(**kwargs):
- return get_vgg(16, **kwargs)
-
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services