Detection and Classification of Dense Tomato Fruits by Integrating Coordinate Attention Mechanism With YOLO Model

Abstract

Real-time detection of objects is one of the important tasks of computer vision applications such as agriculture, surveillance, self-driving cars, etc. The fruit target detection rate based on traditional approaches is low due to the complex background, substantial texture interference, partial occlusion of fruits, etc. This chapter proposes an improved YOLOv5 model to detect and classify the dense tomatoes by adding the coordinate attention mechanism and bidirectional pyramid network. The coordinate attention mechanism is used to detect and classify the dense tomatoes, and bidirectional pyramid network is used to detect the tomatoes at different scales. The proposed model produces good results in detecting the small dense tomatoes with an accuracy of 87.4%.