안녕하세요, HELLO
오늘은 DeepLearning.AI에서 진행하는 앤드류 응(Andrew Ng) 교수님의 딥러닝 전문화의 네 번째 과정인 "Convolutional Neural Networks"을 정리하려고 합니다.
"Convolutional Neural Networks"의 강의를 통해 '자율 주행, 얼굴 인식, 방사선 이미지 인식' 등을 이해하고, CNN 모델에 대해서 배우게 됩니다. 강의는 아래와 같이 구성되어 있습니다.
~ Foundations of Convolutional Neural Networks
~ Deep Convolutional Models: Case Studies
~ Object Detection
~ Special Applications: Face recognition & Neural Style Transfer
"Convolutional Neural Networks" (Andrew Ng)의 3주차 "Image segmentation Unet v2" 실습 내용입니다.
By the time you finish this notebook, you'll be able to:
- Build your own U-Net
- Explain the difference between a regular CNN and a U-net
- Implement semantic image segmentation on the CARLA self-driving car dataset
- Apply sparse categorical crossentropy for pixelwise prediction
CHAPTER 1. 'Packages'
CHAPTER 2. 'Load and split the Data'
CHAPTER 3. 'U-Net'
CHAPTER 4. 'Train the model'
CHAPTER 1. 'Packages'
import tensorflow as tf
import numpy as np
from tensorflow.keras.layers import Input
from tensorflow.keras.layers import Conv2D
from tensorflow.keras.layers import MaxPooling2D
from tensorflow.keras.layers import Dropout
from tensorflow.keras.layers import Conv2DTranspose
from tensorflow.keras.layers import concatenate
from test_utils import summary, comparator
CHAPTER 2. 'Load and split the Data'
import os
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import imageio
import matplotlib.pyplot as plt
%matplotlib inline
path = ''
image_path = os.path.join(path, './data/CameraRGB/')
mask_path = os.path.join(path, './data/CameraMask/')
image_list = os.listdir(image_path)
mask_list = os.listdir(mask_path)
image_list = [image_path+i for i in image_list]
mask_list = [mask_path+i for i in mask_list]
□ Check out the some of the unmasked and masked images from the dataset
After you are done exploring, revert back to `N=2`. Otherwise the autograder will throw a `list index out of range` error.
N = 2
img = imageio.imread(image_list[N])
mask = imageio.imread(mask_list[N])
#mask = np.array([max(mask[i, j]) for i in range(mask.shape[0]) for j in range(mask.shape[1])]).reshape(img.shape[0], img.shape[1])
fig, arr = plt.subplots(1, 2, figsize=(14, 10))
arr[0].imshow(img)
arr[0].set_title('Image')
arr[1].imshow(mask[:, :, 0])
arr[1].set_title('Segmentation')
□ Split Your Dataset into Unmasked and Masked Images
image_list_ds = tf.data.Dataset.list_files(image_list, shuffle=False)
mask_list_ds = tf.data.Dataset.list_files(mask_list, shuffle=False)
for path in zip(image_list_ds.take(3), mask_list_ds.take(3)):
print(path)
image_filenames = tf.constant(image_list)
masks_filenames = tf.constant(mask_list)
dataset = tf.data.Dataset.from_tensor_slices((image_filenames, masks_filenames))
for image, mask in dataset.take(1):
print(image)
print(mask)
□ Preprocess Your Data
Normally, you normalize your image values by dividing them by 255. This sets them between 0 and 1. However, using tf.image.convert_image_dtype with tf.float32 sets them between 0 and 1 for you, so there's no need to further divide them by 255.
def process_path(image_path, mask_path):
img = tf.io.read_file(image_path)
img = tf.image.decode_png(img, channels=3)
img = tf.image.convert_image_dtype(img, tf.float32)
mask = tf.io.read_file(mask_path)
mask = tf.image.decode_png(mask, channels=3)
mask = tf.math.reduce_max(mask, axis=-1, keepdims=True)
return img, mask
def preprocess(image, mask):
input_image = tf.image.resize(image, (96, 128), method='nearest')
input_mask = tf.image.resize(mask, (96, 128), method='nearest')
return input_image, input_mask
image_ds = dataset.map(process_path)
processed_image_ds = image_ds.map(preprocess)
CHAPTER 3. 'U-Net'
U-Net, named for its U-shape, was originally created in 2015 for tumor detection, but in the years since has become a very popular choice for other semantic segmentation tasks.
U-Net builds on a previous architecture called the Fully Convolutional Network, or FCN, which replaces the dense layers found in a typical CNN with a transposed convolution layer that upsamples the feature map back to the size of the original input image, while preserving the spatial information. This is necessary because the dense layers destroy spatial information (the "where" of the image), which is an essential part of image segmentation tasks. An added bonus of using transpose convolutions is that the input size no longer needs to be fixed, as it does when dense layers are used.
Unfortunately, the final feature layer of the FCN suffers from information loss due to downsampling too much. It then becomes difficult to upsample after so much information has been lost, causing an output that looks rough.
U-Net improves on the FCN, using a somewhat similar design, but differing in some important ways. Instead of one transposed convolution at the end of the network, it uses a matching number of convolutions for downsampling the input image to a feature map, and transposed convolutions for upsampling those maps back up to the original input image size. It also adds skip connections, to retain information that would otherwise become lost during encoding. Skip connections send information to every upsampling layer in the decoder from the corresponding downsampling layer in the encoder, capturing finer information while also keeping computation low. These help prevent information loss, as well as model overfitting.
□ Model Details
1. Contracting path (Encoder containing downsampling steps):
Images are first fed through several convolutional layers which reduce height and width, while growing the number of channels.
The contracting path follows a regular CNN architecture, with convolutional layers, their activations, and pooling layers to downsample the image and extract its features. In detail, it consists of the repeated application of two 3 x 3 unpadded convolutions, each followed by a rectified linear unit (ReLU) and a 2 x 2 max pooling operation with stride 2 for downsampling. At each downsampling step, the number of feature channels is doubled.
2. Crop function: This step crops the image from the contracting path and concatenates it to the current image on the expanding path to create a skip connection.
3. Expanding path (Decoder containing upsampling steps):
The expanding path performs the opposite operation of the contracting path, growing the image back to its original size, while shrinking the channels gradually.
In detail, each step in the expanding path upsamples the feature map, followed by a 2 x 2 convolution (the transposed convolution). This transposed convolution halves the number of feature channels, while growing the height and width of the image.
Next is a concatenation with the correspondingly cropped feature map from the contracting path, and two 3 x 3 convolutions, each followed by a ReLU. You need to perform cropping to handle the loss of border pixels in every convolution.
4. Final Feature Mapping Block: In the final layer, a 1x1 convolution is used to map each 64-component feature vector to the desired number of classes. The channel dimensions from the previous layer correspond to the number of filters used, so when you use 1x1 convolutions, you can transform that dimension by choosing an appropriate number of 1x1 filters. When this idea is applied to the last layer, you can reduce the channel dimensions to have one layer per class.
The U-Net network has 23 convolutional layers in total.
Important Note:
The figures shown in the assignment for the U-Net architecture depict the layer dimensions and filter sizes as per the original paper on U-Net with smaller images. However, due to computational constraints for this assignment, you will code only half of those filters. The purpose of showing you the original dimensions is to give you the flavour of the original U-Net architecture. The important takeaway is that you multiply by 2 the number of filters used in the previous step. The notebook includes all of the necessary instructions and hints to help you code the U-Net architecture needed for this assignment.
□ Encoder (Downsampling Block)
The encoder is a stack of various conv_blocks:
Each conv_block() is composed of 2 Conv2D layers with ReLU activations. We will apply Dropout, and MaxPooling2D to some conv_blocks, as you will verify in the following sections, specifically to the last two blocks of the downsampling.
The function will return two tensors:
- next_layer: That will go into the next block.
- skip_connection: That will go into the corresponding decoding block.
Note: If max_pooling=True, the next_layer will be the output of the MaxPooling2D layer, but the skip_connection will be the output of the previously applied layer(Conv2D or Dropout, depending on the case). Else, both results will be identical.
# UNQ_C1
# GRADED FUNCTION: conv_block
def conv_block(inputs=None, n_filters=32, dropout_prob=0, max_pooling=True):
"""
Convolutional downsampling block
Arguments:
inputs -- Input tensor
n_filters -- Number of filters for the convolutional layers
dropout_prob -- Dropout probability
max_pooling -- Use MaxPooling2D to reduce the spatial dimensions of the output volume
Returns:
next_layer, skip_connection -- Next layer and skip connection outputs
"""
### START CODE HERE
conv = Conv2D(n_filters, # Number of filters
3, # Kernel size
activation='relu',
padding='same',
kernel_initializer='he_normal')(inputs)
conv = Conv2D(n_filters, # Number of filters
3, # Kernel size
activation='relu',
padding='same',
kernel_initializer='he_normal')(conv)
### END CODE HERE
# if dropout_prob > 0 add a dropout layer, with the variable dropout_prob as parameter
if dropout_prob > 0:
### START CODE HERE
conv = Dropout(dropout_prob)(conv)
### END CODE HERE
# if max_pooling is True add a MaxPooling2D with 2x2 pool_size
if max_pooling:
### START CODE HERE
next_layer = MaxPooling2D()(conv)
### END CODE HERE
else:
next_layer = conv
skip_connection = conv
return next_layer, skip_connection
input_size=(96, 128, 3)
n_filters = 32
inputs = Input(input_size)
cblock1 = conv_block(inputs, n_filters * 1)
model1 = tf.keras.Model(inputs=inputs, outputs=cblock1)
output1 = [['InputLayer', [(None, 96, 128, 3)], 0],
['Conv2D', (None, 96, 128, 32), 896, 'same', 'relu', 'HeNormal'],
['Conv2D', (None, 96, 128, 32), 9248, 'same', 'relu', 'HeNormal'],
['MaxPooling2D', (None, 48, 64, 32), 0, (2, 2)]]
print('Block 1:')
for layer in summary(model1):
print(layer)
comparator(summary(model1), output1)
inputs = Input(input_size)
cblock1 = conv_block(inputs, n_filters * 32, dropout_prob=0.1, max_pooling=True)
model2 = tf.keras.Model(inputs=inputs, outputs=cblock1)
output2 = [['InputLayer', [(None, 96, 128, 3)], 0],
['Conv2D', (None, 96, 128, 1024), 28672, 'same', 'relu', 'HeNormal'],
['Conv2D', (None, 96, 128, 1024), 9438208, 'same', 'relu', 'HeNormal'],
['Dropout', (None, 96, 128, 1024), 0, 0.1],
['MaxPooling2D', (None, 48, 64, 1024), 0, (2, 2)]]
print('\nBlock 2:')
for layer in summary(model2):
print(layer)
comparator(summary(model2), output2)
□ Decoder (Upsampling Block)
The decoder, or upsampling block, upsamples the features back to the original image size. At each upsampling level, you'll take the output of the corresponding encoder block and concatenate it before feeding to the next decoder block.
There are two new components in the decoder: up and merge. These are the transpose convolution and the skip connections. In addition, there are two more convolutional layers set to the same parameters as in the encoder.
Here you'll encounter the Conv2DTranspose layer, which performs the inverse of the Conv2D layer. You can read more about it here.
# UNQ_C2
# GRADED FUNCTION: upsampling_block
def upsampling_block(expansive_input, contractive_input, n_filters=32):
"""
Convolutional upsampling block
Arguments:
expansive_input -- Input tensor from previous layer
contractive_input -- Input tensor from previous skip layer
n_filters -- Number of filters for the convolutional layers
Returns:
conv -- Tensor output
"""
### START CODE HERE
up = Conv2DTranspose(
n_filters, # number of filters
(3, 3), # Kernel size
strides=2,
padding='same')(expansive_input)
# Merge the previous output and the contractive_input
merge = concatenate([up, contractive_input], axis=3)
conv = Conv2D(n_filters, # Number of filters
(3, 3), # Kernel size
activation='relu',
padding='same',
kernel_initializer='he_normal')(merge)
conv = Conv2D(n_filters, # Number of filters
(3, 3), # Kernel size
activation='relu',
padding='same',
kernel_initializer='he_normal')(conv)
### END CODE HERE
return conv
input_size1=(12, 16, 256)
input_size2 = (24, 32, 128)
n_filters = 32
expansive_inputs = Input(input_size1)
contractive_inputs = Input(input_size2)
cblock1 = upsampling_block(expansive_inputs, contractive_inputs, n_filters * 1)
model1 = tf.keras.Model(inputs=[expansive_inputs, contractive_inputs], outputs=cblock1)
output1 = [['InputLayer', [(None, 12, 16, 256)], 0],
['Conv2DTranspose', (None, 24, 32, 32), 73760],
['InputLayer', [(None, 24, 32, 128)], 0],
['Concatenate', (None, 24, 32, 160), 0],
['Conv2D', (None, 24, 32, 32), 46112, 'same', 'relu', 'HeNormal'],
['Conv2D', (None, 24, 32, 32), 9248, 'same', 'relu', 'HeNormal']]
print('Block 1:')
for layer in summary(model1):
print(layer)
comparator(summary(model1), output1)
□ Build the Model
This is where you'll put it all together, by chaining the encoder, bottleneck, and decoder! You'll need to specify the number of output channels, which for this particular set would be 23. That's because there are 23 possible labels for each pixel in this self-driving car dataset.
# UNQ_C3
# GRADED FUNCTION: unet_model
def unet_model(input_size=(96, 128, 3), n_filters=32, n_classes=23):
"""
Unet model
Arguments:
input_size -- Input shape
n_filters -- Number of filters for the convolutional layers
n_classes -- Number of output classes
Returns:
model -- tf.keras.Model
"""
inputs = Input(input_size)
# Contracting Path (encoding)
# Add a conv_block with the inputs of the unet_ model and n_filters
### START CODE HERE
cblock1 = conv_block(inputs, n_filters)
# Chain the first element of the output of each block to be the input of the next conv_block.
# Double the number of filters at each new step
cblock2 = conv_block(cblock1[0], n_filters*2)
cblock3 = conv_block(cblock2[0], n_filters*4)
cblock4 = conv_block(cblock3[0], n_filters*8, dropout_prob=0.3) # Include a dropout_prob of 0.3 for this layer
# Include a dropout_prob of 0.3 for this layer, and avoid the max_pooling layer
cblock5 = conv_block(cblock4[0], n_filters*16, dropout_prob=0.3, max_pooling=False)
### END CODE HERE
# Expanding Path (decoding)
# Add the first upsampling_block.
# Use the cblock5[0] as expansive_input and cblock4[1] as contractive_input and n_filters * 8
### START CODE HERE
ublock6 = upsampling_block(cblock5[0], cblock4[1], n_filters*8)
# Chain the output of the previous block as expansive_input and the corresponding contractive block output.
# Note that you must use the second element of the contractive block i.e before the maxpooling layer.
# At each step, use half the number of filters of the previous block
ublock7 = upsampling_block(ublock6, cblock3[1], n_filters*4)
ublock8 = upsampling_block(ublock7, cblock2[1], n_filters*2)
ublock9 = upsampling_block(ublock8, cblock1[1], n_filters)
### END CODE HERE
conv9 = Conv2D(n_filters,
3,
activation='relu',
padding='same',
kernel_initializer='he_normal')(ublock9)
# Add a Conv2D layer with n_classes filter, kernel size of 1 and a 'same' padding
### START CODE HERE
conv10 = Conv2D(n_classes, (1,1), padding='same')(conv9)
### END CODE HERE
model = tf.keras.Model(inputs=inputs, outputs=conv10)
return model
import outputs
img_height = 96
img_width = 128
num_channels = 3
unet = unet_model((img_height, img_width, num_channels))
comparator(summary(unet), outputs.unet_model_output)
□ Set Model Dimensions
img_height = 96
img_width = 128
num_channels = 3
unet = unet_model((img_height, img_width, num_channels))
□ Loss Function
In semantic segmentation, you need as many masks as you have object classes. In the dataset you're using, each pixel in every mask has been assigned a single integer probability that it belongs to a certain class, from 0 to num_classes-1. The correct class is the layer with the higher probability.
This is different from categorical crossentropy, where the labels should be one-hot encoded (just 0s and 1s). Here, you'll use sparse categorical crossentropy as your loss function, to perform pixel-wise multiclass prediction. Sparse categorical crossentropy is more efficient than other loss functions when you're dealing with lots of classes.
unet.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
□ Dataset Handling
Below, define a function that allows you to display both an input image, and its ground truth: the true mask. The true mask is what your trained model output is aiming to get as close to as possible.
def display(display_list):
plt.figure(figsize=(15, 15))
title = ['Input Image', 'True Mask', 'Predicted Mask']
for i in range(len(display_list)):
plt.subplot(1, len(display_list), i+1)
plt.title(title[i])
plt.imshow(tf.keras.preprocessing.image.array_to_img(display_list[i]))
plt.axis('off')
plt.show()
for image, mask in image_ds.take(1):
sample_image, sample_mask = image, mask
print(mask.shape)
display([sample_image, sample_mask])
CHAPTER 4. 'Train the model'
EPOCHS = 40
VAL_SUBSPLITS = 5
BUFFER_SIZE = 500
BATCH_SIZE = 32
processed_image_ds.batch(BATCH_SIZE)
train_dataset = processed_image_ds.cache().shuffle(BUFFER_SIZE).batch(BATCH_SIZE)
print(processed_image_ds.element_spec)
model_history = unet.fit(train_dataset, epochs=EPOCHS)
□ Create Predicted Masks
Now, define a function that uses tf.argmax in the axis of the number of classes to return the index with the largest value and merge the prediction into a single image:
def create_mask(pred_mask):
pred_mask = tf.argmax(pred_mask, axis=-1)
pred_mask = pred_mask[..., tf.newaxis]
return pred_mask[0]
□ Plot Model Accuracy
plt.plot(model_history.history["accuracy"])
□ Show Predictions
def show_predictions(dataset=None, num=1):
"""
Displays the first image of each of the num batches
"""
if dataset:
for image, mask in dataset.take(num):
pred_mask = unet.predict(image)
display([image[0], mask[0], create_mask(pred_mask)])
else:
display([sample_image, sample_mask,
create_mask(unet.predict(sample_image[tf.newaxis, ...]))])
show_predictions(train_dataset, 2)
Conclusion
You've come to the end of this assignment. Awesome work creating a state-of-the art model for semantic image segmentation! This is a very important task for self-driving cars to get right. Elon Musk will surely be knocking down your door at any moment. ;)
What you should remember:
- Semantic image segmentation predicts a label for every single pixel in an image
- U-Net uses an equal number of convolutional blocks and transposed convolutions for downsampling and upsampling
- Skip connections are used to prevent border pixel information loss and overfitting in U-Net
■ 마무리
"Convolutional Neural Networks" (Andrew Ng)의 3주차 "Image segmentation Unet v2" 실습에 대해서 정리해봤습니다.
그럼 오늘 하루도 즐거운 나날 되길 기도하겠습니다
좋아요와 댓글 부탁드립니다 :)
감사합니다.
'COURSERA' 카테고리의 다른 글
week 4_Neural Style Transfer (Andrew Ng) (0) | 2022.03.13 |
---|---|
week 4_Face recognition (Andrew Ng) (0) | 2022.03.13 |
week 3_Autonomous driving application Car detection 실습 (Andrew Ng) (0) | 2022.03.13 |
week 3_Object Detection 연습 문제 (Andrew Ng) (0) | 2022.03.13 |
week 3_Object Detection (Andrew Ng) (0) | 2022.03.13 |
댓글