To achieve high-precision Ulva prolifera semantic segmentation from remote sensing imagery and address issues such as boundary fragmentation, contour dilation, and missed segmentation of scattered patches under complex marine backgrounds, this paper proposes an improved SegFormer-based network termed ECAB-SegFormer. The proposed method enhances near-infrared feature representation and boundary perception by embedding an Efficient Channel Attention (ECA) module into shallow features and introducing a boundary supervision branch. Experimental results on the HYU dataset demonstrate that the proposed method achieves consistent improvements over classical baseline models and further outperforms several representative modern strong segmentation baselines. Compared with advanced methods such as DeepLabV3+, Swin-Unet, and Gated-SCNN, the proposed model achieves maximum improvements of 2.77%, 5.80%, and 4.26(pixel) in mIoU, BFScore, and Hausdorff Distance (HD), respectively, while also obtaining superior Precision and F1 Scores. These results demonstrate significant advantages in both regional segmentation accuracy and boundary localization quality, validating the effectiveness, robustness, and practical potential of the proposed method for Ulva prolifera semantic segmentation in remote sensing applications.
Liang et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: