This project aims at wound area segmentation from natural images in clinical settings. The architectures tested so far includes: U-Net, MobileNetV2, Mask-RCNN, SegNet, VGG16.
Wang, C., Anisuzzaman, D.M., Williamson, V. et al. Fully automatic wound segmentation with deep convolutional neural networks. Sci Rep 10, 21897 (2020). https://doi.org/10.1038/s41598-020-78799-w
The training dataset is built by our lab and collaboration clinic, Advancing the Zenith of Healthcare (AZH) Wound and Vascular Center. With their permission, we are sharing this dataset (./data/wound_dataset/) publicly. This dataset was fully annotated by wound professionals and preprocessed with cropping and zero-padding. We plan to publish the raw images and annotations as a segmentation challenge of MICCAI 2021.
python3 train.py
python3 predict.py