Polyp-PVT: Polyp Segmentation with Pyramid Vision Transformers

• DOI: 10.26599/air.2023.9150015
• Publication Date: 2023-06-30T08:37:16Z
• Authors: Bo Dong; Wenhai Wang; Deng-Ping Fan; Jinpeng Li; Huazhu Fu; Ling Shao
• Keywords: Pyramid (geometry)

Most polyp segmentation methods use convolutional neural networks (CNNs) as their backbone, leading to two key issues when exchanging information between the encoder and decoder: (1) taking into account differences in contribution different-level features, (2) designing an effective mechanism for fusing these features. Unlike existing CNN-based methods, we adopt a transformer encoder, which learns more powerful robust representations. In addition, considering image acquisition influence elusive properties of polyps, introduce three standard modules, including cascaded fusion module (CFM), camouflage identification (CIM), similarity aggregation (SAM). Among these, CFM is used collect semantic location polyps from high-level features; CIM applied capture disguised low-level SAM extends pixel features area with position entire area, thereby effectively cross-level The proposed model, named Polyp-PVT, suppresses noises significantly improves expressive capabilities. Extensive experiments on five widely adopted datasets show that model various challenging situations (e.g., appearance changes, small objects, rotation) than representative methods. available at <a ext-link-type="uri" href="https://github.com/DengPingFan/Polyp-PVT">https://github.com/DengPingFan/Polyp-PVT</a>.