Back to Education
Deep Learning Specialization · Convolutional Neural Networks

U-Net Image Segmentation for Autonomous Driving

A semantic segmentation model for autonomous-driving scenes, using a U-Net encoder-decoder with skip connections to classify each pixel.

Semantic segmentationU-NetEncoder-decoderSkip connectionsSparse categorical crossentropy
GitHub

What this project is about

U-Net segmentation changes the output from one label per image to one label per pixel. The encoder compresses visual context, the decoder restores spatial resolution, and skip connections preserve detail.

Implementation focus

This milestone prepared me for medical segmentation and autonomous-driving perception: dense prediction requires architecture choices that protect spatial information.

Key results and artifacts

  • Input size: 96×128×3
  • Classes: 23
  • Model parameters: ~8.64M
  • Training accuracy reached about 0.9849

What I learned

This project changed the task from image-level labels to dense perception: every pixel becomes a prediction.

Milestone deep dive

What this project proves in my learning path.

A semantic segmentation model for autonomous-driving scenes, using a U-Net encoder-decoder with skip connections to classify each pixel. The milestone is included because it shows a complete learning unit: a problem, a data representation, an implementation path, measurable outputs, and a lesson that later connects to larger AI systems.

Input layerpaired images and masks
Model layerencoder-decoder CNN with skip connections
Evaluation layertests, metrics, plots, or qualitative output
Artifact layerpixel-level class masks
Problem framing

What I was trying to solve

This page is not just a lab description. It explains how I transformed the assignment idea into an engineering story: what the model receives, what it optimizes, what output it should produce, and how the result can be inspected.

Implementation

What I coded or assembled

The implementation focus was the mechanism behind the result: functions, model blocks, training logic, preprocessing, evaluation, and the small technical details that make the notebook work rather than only look correct.

Concepts

Core concepts

Semantic segmentationU-NetEncoder-decoderSkip connectionsSparse categorical crossentropy

These tags show the technical ideas this milestone added to my toolkit.

Learning result

What changed after it

This project changed the task from image-level labels to dense perception: every pixel becomes a prediction.

Important outputs and evidence

  • Input size: 96×128×3
  • Classes: 23
  • Model parameters: ~8.64M
  • Training accuracy reached about 0.9849

How I would explain it in an interview

I would describe this milestone by starting from the data representation, then explain the model or algorithm, then show what was measured. The important part is not only the final number; it is the ability to reason about why the method works, what assumptions it makes, and what limitations should be improved next.