Machine Learning for Image Segmentation — HIDA Lecture

This lecture introduces Machine Learning for Image Segmentation using a hands-on, guided Jupyter Notebook designed for Google Colab.

The notebook provides a step-by-step introduction to modern segmentation methods and walks through both foreground–background segmentation using a U-Net and instance segmentation using CellPose.
You will explore how models learn to separate cells in microscopy images and how these techniques can be applied to quantitative biological analysis such as cell counting and morphology statistics.

---

How to Use This Notebook

Access via Google Colab

You can open the notebook directly in Google Colab by clicking the badge below:

---

Setting up Google Colab

Requirements

• A Google account with access to Google Drive.

Important Notes

• Colab resources (CPU/GPU/RAM) are not guaranteed and may become temporarily unavailable.
• Idle sessions longer than 90 minutes or total runtimes exceeding 12 hours will disconnect.
• Unsaved work (e.g., model weights) will be lost when the session ends.

Enable GPU Acceleration
Before running the notebook:

1. In Colab, go to the Runtime menu.
2. Select Change runtime type.
3. Under Hardware accelerator, choose GPU, then click Save.

---

Manual Repository Setup (Alternative)

If you prefer to run the notebook locally or in your own Jupyter environment:

git clone https://github.com/Kainmueller-Lab/HIDA_lecture_image_segmentation.git