IRACLE TECHNOLOGIES SDN.BHD. Software engineering development in AI Enhances Recirculating Aquaculture System.

AI RAS (Artificial Intelligence in Recirculating Aquaculture Systems) is the cutting-edge integration of machine learning and automation into land-based fish farming.

While RAS itself is a closed-loop system that recycles up to 99% of its water, AI acts as the "brain" of the operation, managing the extreme complexity and tight margins for error that come with high-density fish farming.

Stereo camera systems are foundational tools in precision aquaculture, providing 3D spatial data that 2D cameras cannot. By using two synchronized lenses to capture overlapping images, these systems enable non-invasive, high-precision monitoring of both biological and structural components of fish farms

Core Applications
-Biomass & Size Estimation: Stereo cameras are primarily used to measure fish fork length and width without physical handling. These 3D measurements are converted into weight estimates with high accuracy (often <1–5% error), critical for optimizing feeding and harvest timing.

-Behavioral Analysis: They track 3D swimming speed, trajectory, and social interactions. Behavioral anomalies, such as changes in swimming depth or speed, serve as early indicators of stress, disease, or hypoxia.

-Structural Inspection: Mounted on ROVs or fixed points, stereo cameras perform detailed 3D inspections of containment nets, mooring lines, and brackets. They can identify and quantify defects like mesh tears, corrosion, or biofouling accumulation.

-Feeding Management: Real-time 3D feedback monitors pellet distribution and fish feeding intensity, allowing operators to adjust feed delivery to minimize waste and environmental runoff.

Technical Implementation System Configuration: Standard setups use a side-by-side (binocular) orientation. A "short baseline" (small distance between lenses) is often used to keep systems compact and reduce light refraction errors in close-range measurements. Deep Learning Integration: Modern systems utilize AI for automated detection (e.g., YOLOv11) and 3D reconstruction. "Post-detection fusion" is the most common method, where 2D key points (snout, tail) are detected in both images and then triangulated to 3D coordinates.

A preliminary application and its error estimates of a simple stereo-camera measure system for the fish farming.Highly sensitive underwater video system for use in turbid aquaculture ponds

AI photogrammetry software in aquaculture

AI photogrammetry in aquaculture uses computer vision and machine learning to create 3D models and extract metrics (like size, weight, and health) from images and videos of aquatic organisms. This non-invasive method provides data for tasks that were traditionally manual, time-consuming, and often stressful for the fish.

the AI RAS aquaculture environment (water movement, light variations, fish occlusion), having a large dataset like 9,000 images is generally an asset. It provides the necessary data diversity and redundancy to produce reliable and accurate 3D models or train effective AI vision systems. The focus should be on image quality, appropriate angles, and sufficient overlap rather than just a raw number, but 9,000 images is a very comprehensive starting point.

AI data analyst and data science

AI data analysts in fish farming use machine learning and advanced analytics to optimize farm operations, enhance efficiency, and ensure sustainability. They translate vast amounts of sensor and visual data into actionable insights for farmers, moving the industry from traditional, manual methods to data-driven decision-making.