GPF-Net: Gated Progressive Fusion Learning for Polyp Re-Identification
This work addresses polyp matching for colorectal cancer diagnosis, but it is incremental as it builds on existing re-identification methods with a specialized fusion strategy.
The paper tackles the problem of polyp re-identification in colonoscopy images, where coarse high-level features lead to poor results for small objects, and proposes a gated progressive fusion network that selectively fuses multi-level features, achieving improved performance over state-of-the-art unimodal models on standard benchmarks.
Colonoscopic Polyp Re-Identification aims to match the same polyp from a large gallery with images from different views taken using different cameras, which plays an important role in the prevention and treatment of colorectal cancer in computer-aided diagnosis. However, the coarse resolution of high-level features of a specific polyp often leads to inferior results for small objects where detailed information is important. To address this challenge, we propose a novel architecture, named Gated Progressive Fusion network, to selectively fuse features from multiple levels using gates in a fully connected way for polyp ReID. On the basis of it, a gated progressive fusion strategy is introduced to achieve layer-wise refinement of semantic information through multi-level feature interactions. Experiments on standard benchmarks show the benefits of the multimodal setting over state-of-the-art unimodal ReID models, especially when combined with the specialized multimodal fusion strategy.