Deep ContourFLow

In This repository you can find the code for both: unsupervised Deep-ContourFlow and One shot learning Deep-ContourFlow.

if you use the the code please cite the following paper:

@misc{habis2024deepcontourflowadvancingactive,
      title={Deep ContourFlow: Advancing Active Contours with Deep Learning}, 
      author={Antoine Habis and Vannary Meas-Yedid and Elsa Angelini and Jean-Christophe Olivo-Marin},
      year={2024},
      eprint={2407.10696},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2407.10696}, 
}

To use this repository please first install torch-contour:

$pip install torch_contour>=1.3.0

Unsupervised Deep ContourFLow:

To use Unsupervised DCF just add your image in images_test_unsupervised_dcf and run the algorithm using the notebook: unsupervised_dcf.ipynb

One shot learning: Application on Dilated Tubules

To replicate the results in the paper ... follow the steps:

0. Download AIDPATH kidney dataset:

1. Register to AIDPATHDB using the link https://mitel.dimi.uniud.it/aidpath-db/app/login.php
2. Create a folder where data will be saved and write its path in path_data inside the config file. Donwload the kidney dataset of AIDPATHDB and put the images in a subfolder called slides.

Manual annotations are available in the github in the generate_annotations folder.

1. Extract images:

cd ./generate_annotations
python extract_images.py

This code uses annotations.csv and the slides downloaded in #0 in folder slides to extract all the patches centered on each dilated tubule and "false dilated tubule".

The code will create in your data folder:

1. A subfolder 'masks' with the ground truth masks of the 2 class: "False" and "True" dilated tubule.
2. A subfolder 'images' with the correponding images.
3. A file contour_init.npy that contains the initial contour corresponding to the detection of lumen for each patch.

2. Run and get results:

cd ./scores
python compute_scores.py

This code take 10 random dilated tubule in each slide and fit the One shot learning DCF and predict for all other object in the slide. It creates a csv file containing all the scores.