← Back to curriculum

Module 5 — Segmentation & instance masks

Welcome to Module 5

Dense prediction vs detection, encoder–decoder intuition, what Module 5 covers, and dataset prep for segmentation.

~25 min read + exercises

Welcome to Module 5 — segmentation & instance masks

Before we begin

Module 4 drew boxes around objects. Module 5 assigns labels per pixel — semantic, instance, and panoptic segmentation.

You will study U-Net, Mask R-CNN, losses and metrics, and train a pet segmentation model.


What Module 5 covers

LessonTopic
1Semantic segmentation & U-Net
2Instance segmentation & Mask R-CNN
3Losses & mIoU
Quiz20 MCQs
ProjectU-Net pet segmentation

Prerequisites: Module 3 CNN fundamentals, Module 4 detection basics (boxes, IoU).


What's next

Lesson 1 — Semantic segmentation & U-Net