Hierarchical Neural Cellular Automata for Lightweight Microscopy Image Classification (hNCA)

Overview

This project is designed for training and evaluating hNCA model using specified microscopy datasets and folds. It leverages Conda for environment management.

Setup

1. Create Conda Environment

To set up the required environment, use the provided env_hnca.yml file:

conda env create -f env_hnca.yml
conda activate env_hnca

Training

To train the model, run the following command:

python3 src/train.py --output #your_path# --train_set #your_dataset# --fold #your_fold# --mode train

Replace:

• #your_path# with the desired output directory.
• #your_dataset# with your dataset name.
• #your_fold# with a fold number (1-5).

Evaluation

To evaluate the trained model, run:

python3 src/train.py --output #your_path# --train_set #your_dataset# --fold #your_fold# --mode eval

Ensure that #your_fold# is one of [1, 2, 3, 4, 5].

Additional Information

• Class-wise distribution of 6 datasets provided.
• Trained checkpoints for fold 1 of each dataset are provided in the folder results/.

Contact

For any questions or issues, feel free to open an issue or reach out! [email protected]