Unique optical design, covering a wavelength range of 200~1700nm, coupled with a self-developed system for fruit data collection. With the help of this system, high-quality big data image information can be obtained.

By utilizing convolutional neural network algorithms, complex background fruit visual synthesis feature detection and analysis can be conducted. This enables customized solutions for different fruits, allowing precise sorting based on indicators such as fruit size, seed size, external damage, and sweetness.

As a result, the challenge of non-destructive detection of sweetness and seed size is successfully resolved.