This repository is dedicated to benchmarking diverse models for the task of Semantic Segmentation in Videofluoroscopic Swallowing Studies (VFSS). The goal is to provide a comprehensive evaluation of various models, including both traditional and state-of-the-art approaches, to identify the most effective techniques for this specific application.
The datasets used for benchmarking consists of both publicly available and private datasets of VFSS images, annotated for semantic segmentation tasks.
- U-Net
- TransUNet
- SwinUNet
The models are evaluated using the following metrics:
- Mean Intersection over Union (Mean IoU)
- Dice Score