EPPS is implemented using the PyTorch framework (1.13 + cu11) with a single GeForce RTX 3090 GPU of 24 GB memory.
We utilized 3 widely used polyp segmentation datasets:
We adopt an 8:1:1 split for training, validation, and testing.
The model can be trained by running train.py, and the results can be tested and visualised by running test.py.
Comparison with other methods on the Kvasir-SEG and CVC-ClinicDB datasets.
Comparison with other methods on the Kvasir-Sessile datasets.