An AIoT-Based Computer Vision System for Post-Larval Shrimp Detection and Counting in Aquaculture
IEEE Access · 2026 · 14 · 81391-81408
DOI: 10.1109/access.2026.3697530 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Yo-Ping Huang und 1 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Science and Technology Council, Taiwan,.
Abstract (Originalsprache)
The development of the Artificial Intelligence of Things (AIoT) has created new opportunities for improving aquaculture management, particularly in shrimp hatchery operations. By integrating computer vision techniques with connected monitoring systems, AIoT technologies enable automated observation of post-larval shrimp populations. In this study, an AIoT-based computer vision system is proposed for detecting and counting post-larval shrimp in aquaculture environments. The proposed framework adopts a YOLO-based detection architecture enhanced with DarknetConv2D to improve feature extraction capability for very small shrimp objects that often appear densely distributed in hatchery conditions. Different from conventional YOLO-based detection approaches, the proposed system emphasizes an integrated AIoT deployment framework that combines lightweight detection, automated counting, edge computing, IoT communication, and remote monitoring. To evaluate the effectiveness of the proposed system, experiments were conducted under two operational scenarios, namely a waterslide scenario representing dynamic water flow conditions and a dam scenario representing stationary water environments. Quantitative results demonstrate that the proposed model achieves 99.1% detection accuracy,94.8% of precision,88.1% of recall, and an F1-score of 91.3%, with an mAP50 of 98.6%. In addition, the model maintains a lightweight architecture with 16.2 M parameters while achieving a high inference speed of 185 FPS, enabling efficient deployment for automated monitoring systems. The proposed system demonstrates strong potential for improving counting accuracy, reducing manual labor, and supporting the development of intelligent aquaculture management systems.
Förderung
- National Science and Technology Council, Taiwan,
Zitieren
APA
Huang, Y., & Surya, I. (2026). An AIoT-Based Computer Vision System for Post-Larval Shrimp Detection and Counting in Aquaculture. IEEE Access, 14, 81391-81408. https://doi.org/10.1109/access.2026.3697530
Vancouver
Huang Y, Surya I. An AIoT-Based Computer Vision System for Post-Larval Shrimp Detection and Counting in Aquaculture. IEEE Access. 2026;14:81391-81408. doi:10.1109/access.2026.3697530
BibTeX
@article{Huang2026,
author = {Huang, Yo-Ping and Surya, Irgi},
title = {An AIoT-Based Computer Vision System for Post-Larval Shrimp Detection and Counting in Aquaculture},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {81391--81408},
doi = {10.1109/access.2026.3697530}
}
EUVIMED