MIDOG 2025 Track 1 Reference Algorithm Docker

This is the initial release (v1.0.0) of the MIDOG 2025 Track 1 reference algorithm docker image.

What is included:

• Pretrained model weights for the FCOS ResNext101 model
• Source code to run inference on histopathology crop out images
• Instructions to create an algorithm for submission on grand-challenge.org

How to get started:

• Clone the repository and download the weigths from this release with the download_weights.py script
• Alternatively, download the checkpoint and save it to resources/FCOS_x101.ckpt manually
• Follow the instructions in the README.md to set up docker and create your own submission

Happy programming!