![[2602.19437] FinSight-Net:A Physics-Aware Decoupled Network with Frequency-Domain Compensation for Underwater Fish Detection in Smart Aquaculture](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19437] FinSight-Net:A Physics-Aware Decoupled Network with Frequency-Domain Compensation for Underwater Fish Detection in Smart Aquaculture
Summary
FinSight-Net introduces a physics-aware framework for underwater fish detection, improving accuracy while reducing computational overhead in smart aquaculture environments.
Why It Matters
This research addresses critical challenges in underwater fish detection, enhancing monitoring capabilities in aquaculture. By integrating physics-based principles, it provides a more efficient and effective solution, which is vital for sustainable practices in marine environments.
Key Takeaways
- FinSight-Net effectively compensates for frequency-specific information loss in underwater environments.
- The framework reduces computational overhead while improving detection accuracy compared to existing models.
- Extensive testing shows FinSight-Net outperforms YOLOv11s in challenging conditions.
- The model is designed for real-time applications in smart aquaculture, enhancing operational efficiency.
- Physics-aware approaches can significantly improve performance in complex detection tasks.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.19437 (cs) [Submitted on 23 Feb 2026] Title:FinSight-Net:A Physics-Aware Decoupled Network with Frequency-Domain Compensation for Underwater Fish Detection in Smart Aquaculture Authors:Jinsong Yang, Zeyuan Hu, Yichen Li, Hong Yu View a PDF of the paper titled FinSight-Net:A Physics-Aware Decoupled Network with Frequency-Domain Compensation for Underwater Fish Detection in Smart Aquaculture, by Jinsong Yang and 3 other authors View PDF HTML (experimental) Abstract:Underwater fish detection (UFD) is a core capability for smart aquaculture and marine ecological monitoring. While recent detectors improve accuracy by stacking feature extractors or introducing heavy attention modules, they often incur substantial computational overhead and, more importantly, neglect the physics that fundamentally limits UFD: wavelength-dependent absorption and turbidity-induced scattering significantly degrade contrast, blur fine structures, and introduce backscattering noise, leading to unreliable localization and recognition. To address these challenges, we propose FinSight-Net, an efficient and physics-aware detection framework tailored for complex aquaculture environments. FinSight-Net introduces a Multi-Scale Decoupled Dual-Stream Processing (MS-DDSP) bottleneck that explicitly targets frequency-specific information loss via heterogeneous convolutional branches, suppressing backscattering artifacts while compensating dist...