SAM Polyp Segmentation Pipeline uses three Vision Transformer encoders (ViT1, ViT2, ViT3) to process polyp images in parallel, each feeding through dedicated adapters into a unified Feature Fusion module. The fused features flow through an Edge Extraction Module to generate edge maps, which are combined with the original fused features via residual addition to create edge-enhanced outputs. This multi-encoder architecture with edge-aware residual connections improves segmentation accuracy by capturing both semantic and boundary information across multiple transformer scales. Fork and customize this pipeline on Diagrams.so to adapt it for your specific endoscopy datasets or integrate it into your medical imaging workflow.