Kaohsiung, Taiwan, Apr 25: Shrimp farming supplies a significant share of the world’s seafood. Yet, many critical decisions: how much to feed, how shrimps live and grow, and when to harvest, are still based on ad hoc experience, occasional sampling, and nighttime patrols. While digital aquaculture tools are advancing globally, many existing platforms are optimized for clear-water, surface-feeding fish species, leaving shrimp farmerswho operate in turbid ponds with bottom-dwelling, primarily nocturnal animalswithout reliable, real-time visibility.
Researchers at National Sun Yat-sen University (NSYSU), led by Professor Ing-Jer Huang, have developed an AI-powered underwater monitoring and decision-support system explicitly designed for shrimp farming in murky pond environments, where conventional visual and sensor-based tools often fall short.
Instead of relying mainly on water-quality sensors or occasional surface checks, the NSYSU system places continuous underwater visual evidence at the center of farm management. Submerged devices operate 24 hours a day, capturing vivid shrimp images even in muddy water. Artificial intelligence algorithms then quantify shrimp size, activity levels, and residual feed, transforming underwater observations into actionable indicators for feeding decisions, growth tracking, and harvest planning.
The technology has progressed beyond experimental trials. It has been deployed and validated at more than 15 aquaculture sites in Taiwan, generating over two years of continuous field data. Comparative assessments show that real-time AI-based estimates of shrimp length and weight differ from manual sampling by only 4.5–6.3%, confirming that the system performs reliably under real farming conditions.
Operational results have also demonstrated clear production benefits. In commercial white-shrimp ponds, farms using AI-assisted feeding optimization reported survival rate increases of up to 20%, feed conversion improvements of around 6%, and overall yield gains of approximately 26%, compared with conventional feeding practices. These gains directly address feed efficiency, cost control, and environmental sustainability, key concerns for shrimp producers worldwide.
By integrating patented underwater observation devices, AI-based biological analysis, edge computing, and cloud-based decision support into a closed-loop system, this work offers a practical pathway to move shrimp aquaculture from experience-driven management to evidence-based, scalable operations that can operate across multiple production cycles.
The research team is now actively seeking collaboration with shrimp farms, feed producers, aquaculture equipment providers, and regional partners, particularly in Southeast Asia, where farming conditions closely resemble those in Taiwan. Potential collaborations include joint pilot deployments, feed optimization studies, system integration, and regional adaptation for large-scale production.
Professor Huang emphasizes that practicality guided the development. “Farmers need systems that work continuously, at night, in muddy water, and under real economic pressure. Our focus has always been on solutions that farmers can rely on every day.”
This project was supported by the National Science and Technology Council, Ministry of Agriculture, CISCO, Chunghwa Telecom, and National Sun Yat-sen University’s Southern Taiwan Growth (STG) initiative, which facilitated industry engagement, pilot implementation, and collaboration with aquaculture partners to help translate research outcomes into real-world applications.
