Traditional crop inspection methods, such as those for grains, beans or fruits, rely heavily on labors. Labors visually examine factors like color uniformity, cracks, size, shape and foreign objects. This process is time-consuming and susceptible to errors due to fatigue, leading to inconsistencies and unstable inspection quality and efficiency.
With advancements in agricultural technology, embedded edge AI systems now offer an efficient crop sorting solution to address the issues of time and error in visual inspection. These systems’ stability, low power consumption, and adaptability to various environments enhance inspection efficiency and ensure quality. In the application, when grains are poured into a hopper, a vibrating feeder moves the grains, and an industrial camera captures images of the grains under controlled lighting. These images are sent to the embedded system, where AI technology analyzes the features and determines quality. Based on this inference, a PLC directs an air pressure device to sort the grains by quality.
The WEBS-21J0-ASL, equipped with the Hailo-8 AI module, delivers 26 TOPS of computational performance, suitable for real-time image inference with accurate results. The system is powered by the Intel Atom x7000RE series SoC, providing sufficient computing capability while operating at low power to suit limited-power agricultural environments. It is designed to operate in temperatures from -20°C to 60°C, withstands dust due to its fanless design, and maintains function in 5-95% relative humidity. The system’s I/O interfaces, including RS-485 for PLC and PCIe for image capture card connection, ensure connectivity with peripheral equipment.
