Optimization Analysis of Soybean Microwave Vacuum Drying Process Based on Improved Yolo V5s Algorithm

Drying technology plays a non-negligible role in video preservation and its transport, but traditional drying technology faces the double challenges of efficiency and product quality. For this reason, this study adopts advanced image processing technology to monitor the soybean drying process in real time and optimise the process parameters through algorithms. In this study, we firstly enhance the feature extraction capability of YOLO v5s algorithm, optimise small target detection, and integrate a multi-task learning framework. In addition, data enhancement techniques are applied to improve model generalisation, adjust the loss function to adapt to multi-task learning needs, and optimise algorithm performance for real-time monitoring. In the experimental part, this study verified the effectiveness of the improved algorithm by comparing the confidence and frame rate before and after the algorithm improvement, as well as testing the drying uniformity and nutrient retention rate. The conclusion shows that the optimised algorithm significantly improves the uniformity of soybean drying and product quality, and the nutrient retention rate reaches more than 85.1% in both cases.