Guides

The guides assume the repository layout described in Project Structure and a CUDA GPU (CPU works but is slow for DINOv2 and the U-Net).

Guide	Goal	Modules involved
Installation	Environment, dependencies, Galaxy10 download	requirements.txt, data_loader.py
Train a Classifier	SimpleCNN or EfficientNet-B0 with the shared training loop	models.py, trainer.py
Run GradCAM	Heat maps for a trained CNN	gradcam.py
Run Segmentation	Otsu masks, U-Net training and inference	segmenter.py, unet.py
Compute Morphometry	CAS / Gini / M20 on a set of images	morphometry.py

Analyses that exist only as notebooks (DINOv2 fine-tuning, fusion, uncertainty, attention morphometry) are listed in Project Structure.

Related pages

▶Installation

Start here

</>API Reference

Module reference