Overcoming annotation bottlenecks in underwater fish segmentation: a robust self-supervised learning approach

作     者：Saleh, Alzayat Sheaves, Marcus Jerry, Dean Azghadi, Mostafa Rahimi 

作者机构：James Cook Univ Coll Sci & Engn Townsville Qld Australia James Cook Univ ARC Res Hub Supercharging Trop Aquaculture Genet S Townsville Qld Australia 

出 版 物：《SIGNAL IMAGE AND VIDEO PROCESSING》 (Signal Image Video Process.)

年 卷 期：2025年第19卷第4期

页      面：1-10页

核心收录：

学科分类：0808[工学-电气工程] 1002[医学-临床医学] 0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 

基　　金：James Cook University Australian Research Training Program (RTP) Scholarship and Food Agility HDR Top-Up Scholarship Australian Research Council through their Industrial Transformation Research Hub program 

主　　题：Computer vision Convolutional neural networks Underwater videos Deep learning Transformer Self-supervised learning 

摘      要：Accurate fish segmentation in underwater videos is challenging due to low visibility, variable lighting, and dynamic backgrounds, making fully-supervised methods that require manual annotation impractical for many applications. This paper introduces a novel self-supervised learning approach for fish segmentation using Deep Learning. Our model, trained without manual annotation, learns robust and generalizable representations by aligning features across augmented views and enforcing spatial-temporal consistency. We demonstrate its effectiveness on three challenging underwater video datasets: DeepFish, Seagrass, and YouTube-VOS, surpassing existing self-supervised methods and achieving segmentation accuracy comparable to fully-supervised methods without the need for costly annotations. Trained on DeepFish, our model exhibits strong generalization, achieving high segmentation accuracy on the unseen Seagrass and YouTube-VOS datasets. Furthermore, our model is computationally efficient due to its parallel processing and efficient anchor sampling technique, making it suitable for real-time applications and potential deployment on edge devices. We present quantitative results using Jaccard Index and Dice coefficient, as well as qualitative comparisons, showcasing the accuracy, robustness, and efficiency of our approach for advancing underwater video analysis.