U-Net architecture in Computer Vision

Introduction

U-Net helps computers find and separate objects in pictures by learning from examples. It is great for tasks like medical image analysis where precise shapes matter.

When you want to identify exact areas of objects in images, like tumors in medical scans.

When you need to separate different parts of a photo, such as roads and buildings in satellite images.

When you want to color or label each pixel in an image for detailed understanding.

When you have limited training data but need accurate image segmentation.

When you want a model that can learn both the big picture and fine details in images.

Syntax

Computer Vision

def unet(input_shape):
    inputs = Input(input_shape)
    # Encoder: Downsampling path
    c1 = Conv2D(64, 3, activation='relu', padding='same')(inputs)
    c1 = Conv2D(64, 3, activation='relu', padding='same')(c1)
    p1 = MaxPooling2D()(c1)

    c2 = Conv2D(128, 3, activation='relu', padding='same')(p1)
    c2 = Conv2D(128, 3, activation='relu', padding='same')(c2)
    p2 = MaxPooling2D()(c2)

    # Bottleneck
    c5 = Conv2D(1024, 3, activation='relu', padding='same')(p2)
    c5 = Conv2D(1024, 3, activation='relu', padding='same')(c5)

    # Decoder: Upsampling path
    u6 = Conv2DTranspose(512, 2, strides=2, padding='same')(c5)
    u6 = concatenate([u6, c2])
    c6 = Conv2D(512, 3, activation='relu', padding='same')(u6)
    c6 = Conv2D(512, 3, activation='relu', padding='same')(c6)

    u7 = Conv2DTranspose(256, 2, strides=2, padding='same')(c6)
    u7 = concatenate([u7, c1])
    c7 = Conv2D(256, 3, activation='relu', padding='same')(u7)
    c7 = Conv2D(256, 3, activation='relu', padding='same')(c7)

    outputs = Conv2D(1, 1, activation='sigmoid')(c7)
    model = Model(inputs, outputs)
    return model

U-Net has two main parts: encoder (downsampling) and decoder (upsampling).

Skip connections copy features from encoder to decoder to keep details.

Examples

Create a U-Net model for 128x128 grayscale images and show its layers.

Computer Vision

inputs = Input((128, 128, 1))
model = unet((128, 128, 1))
model.summary()

Change output layer to predict 3 classes instead of 1 for multi-class segmentation.

Computer Vision

outputs = Conv2D(3, 1, activation='softmax')(c7)

Sample Model

This code builds a small U-Net, trains it on random data for 2 rounds, and shows loss, accuracy, and prediction shape.

Computer Vision

import numpy as np
from tensorflow.keras.layers import Input, Conv2D, MaxPooling2D, Conv2DTranspose, concatenate
from tensorflow.keras.models import Model
from tensorflow.keras.optimizers import Adam

# Define U-Net model
def unet(input_shape):
    inputs = Input(input_shape)
    c1 = Conv2D(16, 3, activation='relu', padding='same')(inputs)
    c1 = Conv2D(16, 3, activation='relu', padding='same')(c1)
    p1 = MaxPooling2D()(c1)

    c2 = Conv2D(32, 3, activation='relu', padding='same')(p1)
    c2 = Conv2D(32, 3, activation='relu', padding='same')(c2)
    p2 = MaxPooling2D()(c2)

    c3 = Conv2D(64, 3, activation='relu', padding='same')(p2)
    c3 = Conv2D(64, 3, activation='relu', padding='same')(c3)

    u4 = Conv2DTranspose(32, 2, strides=2, padding='same')(c3)
    u4 = concatenate([u4, c2])
    c4 = Conv2D(32, 3, activation='relu', padding='same')(u4)
    c4 = Conv2D(32, 3, activation='relu', padding='same')(c4)

    u5 = Conv2DTranspose(16, 2, strides=2, padding='same')(c4)
    u5 = concatenate([u5, c1])
    c5 = Conv2D(16, 3, activation='relu', padding='same')(u5)
    c5 = Conv2D(16, 3, activation='relu', padding='same')(c5)

    outputs = Conv2D(1, 1, activation='sigmoid')(c5)
    model = Model(inputs, outputs)
    return model

# Create model
model = unet((64, 64, 1))
model.compile(optimizer=Adam(), loss='binary_crossentropy', metrics=['accuracy'])

# Create dummy data: 10 images 64x64 with 1 channel
x_train = np.random.rand(10, 64, 64, 1).astype(np.float32)
y_train = (x_train > 0.5).astype(np.float32)  # dummy masks

# Train model for 2 epochs
history = model.fit(x_train, y_train, epochs=2, batch_size=2, verbose=0)

# Predict on one image
pred = model.predict(x_train[:1])
print(f"Loss after 2 epochs: {history.history['loss'][-1]:.4f}")
print(f"Accuracy after 2 epochs: {history.history['accuracy'][-1]:.4f}")
print(f"Prediction shape: {pred.shape}")

Important Notes

U-Net works well even with small datasets because of its design.

Skip connections help keep image details during upsampling.

Output activation is usually sigmoid for binary masks or softmax for multiple classes.

Summary

U-Net is a special model for finding exact shapes in images.

It has two parts: one shrinks the image to learn features, the other grows it back to original size.

Skip connections link these parts to keep details and improve accuracy.

Practice

1. What is the main purpose of the U-Net architecture in computer vision?

easy

A. To compress images for storage

B. To classify images into categories

C. To find exact shapes and boundaries in images

D. To generate new images from noise

Solution

1. Step 1: Understand U-Net's design goal

   U-Net is designed to segment images, meaning it finds exact shapes and boundaries.

2. Step 2: Compare with other tasks

   Classification assigns labels, generation creates images, compression reduces size; these are not U-Net's main tasks.

3. Final Answer:

   To find exact shapes and boundaries in images -> Option C

4. Quick Check:

   U-Net = segmentation [OK]

Hint: U-Net = image segmentation for shapes [OK]

Common Mistakes:

• Confusing segmentation with classification
• Thinking U-Net generates images
• Assuming U-Net compresses images

2. Which of the following correctly describes the structure of U-Net?

easy

A. Two paths: one shrinks and one expands the image size

B. A single path that shrinks the image size

C. A path that only expands the image size

D. Multiple independent paths without connection

Solution

1. Step 1: Recall U-Net architecture parts

   U-Net has an encoder path that shrinks and a decoder path that expands the image size.

2. Step 2: Eliminate incorrect options

   Single shrinking path or only expanding path do not match U-Net. Independent paths without connection are incorrect.

3. Final Answer:

   Two paths: one shrinks and one expands the image size -> Option A

4. Quick Check:

   U-Net = encoder + decoder [OK]

Hint: U-Net has encoder and decoder paths [OK]

Common Mistakes:

• Thinking U-Net only shrinks images
• Believing U-Net only expands images
• Ignoring the connection between paths

3. Consider this simplified U-Net code snippet in Python using PyTorch:

import torch
import torch.nn as nn

class SimpleUNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.down = nn.Conv2d(1, 16, 3, padding=1)
        self.up = nn.ConvTranspose2d(16, 1, 2, stride=2)

    def forward(self, x):
        x1 = self.down(x)  # shrink
        x2 = self.up(x1)   # expand
        return x2

model = SimpleUNet()
input_tensor = torch.randn(1, 1, 64, 64)
output = model(input_tensor)
print(output.shape)

What will be the printed output shape?

medium

A. torch.Size([1, 1, 64, 64])

B. torch.Size([1, 1, 128, 128])

C. torch.Size([1, 16, 64, 64])

D. torch.Size([1, 16, 32, 32])

Solution

1. Step 1: Analyze down convolution output size

   Input is 64x64, Conv2d with kernel=3, padding=1 keeps size 64x64, output channels=16.

2. Step 2: Analyze up convolution output size

   ConvTranspose2d with kernel=2, stride=2 doubles spatial size to 128x128, output channels=1.

3. Step 3: Check actual output shape

   Output shape is (batch=1, channels=1, height=128, width=128).

4. Final Answer:

   torch.Size([1, 1, 128, 128]) -> Option B

5. Quick Check:

   Upsample doubles size = 128x128 [OK]

Hint: ConvTranspose2d with stride 2 doubles size [OK]

Common Mistakes:

• Assuming Conv2d changes size without padding
• Ignoring stride effect in ConvTranspose2d
• Mixing channel numbers in output

4. In a U-Net implementation, a student forgot to add skip connections between encoder and decoder layers. What is the most likely effect?

medium

A. The model will not train at all and cause an error

B. The model will have better accuracy due to fewer connections

C. The model will run faster and produce the same results

D. The model will lose fine details and perform worse on segmentation

Solution

1. Step 1: Understand skip connections role

   Skip connections pass fine details from encoder to decoder to improve segmentation accuracy.

2. Step 2: Effect of missing skip connections

   Without them, decoder lacks detail info, causing worse segmentation results.

3. Final Answer:

   The model will lose fine details and perform worse on segmentation -> Option D

4. Quick Check:

   Skip connections = better detail [OK]

Hint: No skip connections = loss of detail [OK]

Common Mistakes:

• Thinking fewer connections always improve accuracy
• Assuming model will error without skip connections
• Believing speed improves without accuracy loss

5. You want to improve a U-Net model's segmentation on small objects in medical images. Which change is most effective?

hard

A. Increase the depth of the encoder to capture more features

B. Remove skip connections to simplify the model

C. Use larger convolution kernels to cover bigger areas

D. Add more pooling layers to reduce spatial size aggressively

Solution

1. Step 1: Identify challenge with small objects

   Small objects need detailed features and context captured by deeper layers.

2. Step 2: Evaluate options

   Increasing encoder depth helps learn more features. Removing skip connections loses detail. Larger kernels may blur small details. More pooling reduces spatial resolution, losing small object info.

3. Final Answer:

   Increase the depth of the encoder to capture more features -> Option A

4. Quick Check:

   Deeper encoder = better small object features [OK]

Hint: Deeper encoder = better small object detection [OK]

Common Mistakes:

• Removing skip connections harms detail
• Larger kernels blur small features
• More pooling loses small object info