Lightweight Detection Method for Tomato Maturity in Complex Environments Based on PCIA-YOLO

In order to solve the problem of low detection accuracy of tomatoes of different ripeness in the occlusion environment of traditional methods, an improved YOLOv8n model was introduced, which was called PCIA-YOLO. In this study, the PCIA-YOLO model was established to improve feature extraction through partial convolution and self-attention mechanism, and the IoU loss function and adaptive sample selection mechanism were used to enhance the accuracy and performance of tomato ripeness detection. The PCIA-YOLO model performs well in the task of tomato ripeness detection, with higher accuracy and faster convergence speed, and at the same time optimizes the model resources, providing an effective technical solution for agricultural automated harvesting.