Implemented SmoothGrad with Chainer v2

Introduction

In image recognition, it is important to visualize what the classifier focused on and recognized. CNN has proposed a method of propagating the gradient of the loss function to the input image by backpropagation and visualizing the intensity of its absolute value, but there was a problem that it was noisy. SmoothGrad is a very easy way to get a clean visualization by simply adding Gaussian noise to the input image and averaging multiple gradients.

cheetah.jpg

How it is averaged cheetah.gif

Smooth Grad MN099.jpg

Vanilla Grad (conventional method) VanillaMM.jpg

The TensorFlow code and papers can be downloaded from the following. https://tensorflow.github.io/saliency/

I also implemented SmoothGrad to study Chainer v2. The model is a trained VGG16 model. The environment is Windows7 64bit Python 3.5.2 |Anaconda 4.2.0 (64-bit) The version of chainer is '2.0.0'. It does not support GPU.

Implementation

import

smoothgrad.py


import chainer
import chainer.functions as F
from chainer.variable import Variable
from chainer.links import VGG16Layers
from PIL import Image
import numpy as np
import matplotlib.pyplot as plt

config We do it in test mode, but we need to do backpropagation, so Set chainer.config as follows.

smoothgrad.py


chainer.config.train=False
chainer.config.enable_backprop=True

VGG16 model road

Load the VGG16 model. The model is about 500MB, so if it has not been downloaded in advance, it will take some time.

smoothgrad.py


model = VGG16Layers()

Image loading and preprocessing

The image size of VGG is 224x224, so resize it.

smoothgrad.py


image = Image.open("cheetah.png ")
image = image.resize((224,224))

Parameters

Set the number of samplings and noise level. The number of samplings is 100 and the noise level is 20%.

smoothgrad.py


sampleSize = 100
noiseLevel = 0.2 # 20%
sigma = noiseLevel*255.0

Gradient calculation

Due to the memory used, we will do it one by one this time. First, in VGG16, the channel arrangement is BGR, so convert it and subtract the average value. Next, while adding image noise, forward propagation and loss are calculated, and back propagation is used to calculate the gradient. Add the calculated gradient to the list.

smoothgrad.py


gradList = []
for _ in range(sampleSize):
    x = np.asarray(image, dtype=np.float32)
    # RGB to BGR
    x = x[:,:,::-1]
    #Subtract the average
    x -= np.array([103.939, 116.779, 123.68], dtype=np.float32)
    x = x.transpose((2, 0, 1))
    x = x[np.newaxis]
    #Add noise
    x += sigma*np.random.randn(x.shape[0],x.shape[1],x.shape[2],x.shape[3])    
    x = Variable(np.asarray(x))
    #FP and take out the final layer
    y = model(x, layers=['prob'])['prob']
    #BP with the highest forecast label
    t = np.zeros((x.data.shape[0]),dtype=np.int32)
    t[:] = np.argmax(y.data)
    t = Variable(np.asarray(t))
    loss = F.softmax_cross_entropy(y,t)
    loss.backward()
    #Add gradient to list
    grad = np.copy(x.grad)
    gradList.append(grad)
    #Clear the gradient
    model.cleargrads()

Visualization

Take the maximum absolute value for each channel of the gradient and average it over the image.

smoothgrad.py


G = np.array(gradList)
M = np.mean(np.max(np.abs(G),axis=2),axis=0)
M = np.squeeze(M)
plt.imshow(M,"gray")
plt.show()

M049.png

result

I tried to visualize the original image and the map pixel by pixel while increasing the average number of sheets.

1 sample MN000.jpg

2 samples MN001.jpg

3 samples MN003.jpg

10 samples MN009.jpg

20 samples MN019.jpg

30 samples MN029.jpg

50 samples MN049.jpg

75 samples MN074.jpg

100 samples MN099.jpg

If you visualize from one sample without giving noise, you can see a lot of noise as shown below. VanillaMM.jpg

Recommended Posts

Implemented SmoothGrad with Chainer v2
Implemented Conditional GAN with chainer
Seq2Seq (1) with chainer
Use tensorboard with Chainer
Migrating from Chainer v1 to Chainer v2
Implemented word2vec with Theano + Keras
Learn elliptical orbits with Chainer
Implemented SMO with Python + NumPy
Seq2Seq (3) ~ CopyNet Edition ~ with chainer
Use chainer with Jetson TK1
Chainer v1.21 has been released
Neural network starting with Chainer
Image caption generation with Chainer
Deep Embedded Clustering with Chainer 2.0
A little stuck with chainer
Zura with softmax function implemented
[Text classification] I implemented Convolutional Neural Networks for Sentence Classification with Chainer
Getting Started with Python responder v2
Implemented file download with Python + Bottle
Multilayer Perceptron with Chainer: Function Fitting
Try horse racing prediction with Chainer
I implemented Attention Seq2Seq with PyTorch
[Chainer] Learning XOR with multi-layer perceptron
First Anime Face Recognition with Chainer
Various Fine Tuning with Mobilenet v2
Inference works with Chainer 2.0 MNIST sample
Using Chainer with CentOS7 [Environment construction]
Try Common Representation Learning with chainer
Seq2Seq (2) ~ Attention Model edition ~ with chainer
[Deprecated] Chainer v1.24.0 Tutorial for beginners
Implemented inter-frame difference method with OpenCV