This project aims to develop a supervisory model for a segmentation network using multiple instance learning. By labeling patches based on their corresponding original masks it is possible to train a model the classification task which then can be used to generate attention maps | attribution maps, subsequently enabling to create binary masks indicating WHM features. Afterwards, model may overwatch segmentation network for improved results.
In progress ...
In progress ...
Contributions are welcome! If you find any issues or have suggestions for improvements, please open an issue or submit a pull request.
This project is licensed under the MIT License.