Exercise for lecture 4 - Image Segmentation

Learning Objectives

• Explore a basic segmentation pipeline with filtering and thresholding
• Evaluate the quality of the segmentation when a ground truth is available

Preparation

• Accept assignment:
• Clone your student repository (git clone)
• Run uv sync and check everything is correct with uv run hello.py
• Start Jupyter

Exercise

1. Evaluate on images the SNR according to different definitions
2. Tune the parameters of different filters in different noise conditions to achieve a target MSE.

Exercise

Notebook 04-Fossil.ipynb has shown you how to apply a threshold on an image to segment it into a bilevel image. The performance of the threshold was improved by applying a filter that reduced the noise levels in the image. Now you can try this workflow on a different image.

1. Repeat the same analysis to the teeth fossil data (teeth_image.tif)
2. Adjust the threshold and angles to try and see the gut structure better
3. Improve the filters by using more advanced techniques and windows (bonus for non-local means)

In tasks.py you will have to implement to common metrics to evalaute segmentation masks:

• Dice coefficient
• Jaccard index

First, you will have to read about the metrics and implement them to pass the corresponding tests. Then, you will have to implement a basic segmentation pipeline on the dataset teeth_image.tif that achieves a dice_coefficient > 0.95 and a jaccard_index > 0.95