Model Architecture

Grad-TransUNet Architecture Figure from Official Paper

Dependencies

• Python 3.6+
• pip install -r requirements.txt

Dataset

• kvasir dataset was used for training.
• First, you should generate the Grad-Cam Images with this github repositorygradcam pytorch
• you can choose differnet models to generate the heatmaps based on them (in my work resenet was the best model in generating heatmaps!)

Training

• Training process can be started with following command.
  • python main.py --mode train --model_path ./path/to/model --train_path ./path/to/trainset --train_sail_path ./path/to/gradcam-trainset --test_path ./path/to/testset --test_sail_path ./path/to/gradcam-testset

Inference

• After model is trained, inference can be run with following command.
  • python main.py --mode inference --model_path ./path/to/model --image_path ./path/to/image

Other Implementations

• Self Attention CV / The AI Summer
• SOTA Vision / 04RR

References

• [1] TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation
• [2] Automatic segmenting teeth in X-ray images: Trends, a novel data set, benchmarking and future perspectives