What it does
The machine sorts fish based on weight and skin discolouration. We found out that fish grading systems are expensive and are inaccesible to small and medium-scale fish farmers. Our system provides a low-cost, accurate, and efficient solution to this problem.
Your inspiration
The inspiration came from observing the challenges faced by small and medium-sized fish farmers. Traditional manual grading is labour-intensive, slow, and prone to error, leading to inconsistent quality and reduced profits. Existing automated systems are often prohibitively expensive for their needs. We wanted to create an affordable, efficient, and accurate solution to support sustainable aquaculture and help local farmers improve productivity.
How it works
Fish are placed on a conveyor belt one at a time. As a fish moves along, it passes over a weight sensor to measure its mass and under a camera (ESP32-CAM) that captures an image to detect signs of illness, like skin discoloration. An Arduino microcontroller processes this data and activates one of three servo-controlled gates, which gently sweeps the fish into the correct bin based on its weight and health. This automates the entire grading process quickly and accurately.
Design process
Our process began with market surveys to understand user needs, followed by a patent search. We generated multiple concepts, using scoring methods to select the most promising design. The chosen concept was then developed using SolidWorks for CAD modeling and Finite Element Analysis (FEA) to ensure structural integrity. We built a functional prototype using aluminum extrusions and 3D-printed PLA parts, integrating off-the-shelf electronics. The process was iterative, with continuous testing and refinement to improve performance.
How it is different
Unlike expensive, industrial-scale systems, our design is specifically tailored for small to medium-sized fish farmers, making automation affordable. It's unique because it combines two grading parameters: weight and visual health (via AI-powered image recognition). Most affordable systems only sort by weight. Our modular design, built with standard profiles and 3D-printed parts, is also highly repairable and customizable, reducing long-term maintenance costs.
Future plans
We plan to refine the prototype for commercial production. Key next steps include developing a custom PCB to make the electronics more compact, and enhancing the AI model to detect a wider range of diseases. We will also develop a mobile app for IoT integration, allowing farmers to monitor the grading process remotely. Finally, we will improve the design to be fully waterproof for field deployment and explore cost-optimisation for mass production, such as using injection moulding.
Share this page on