Local zoom system for agricultural pest detection and recognition

The ability to detect and recognize insect pests is of great importance for the output and quality of agricultural production. Computer vision is widely used in pest image detection and recognition. However, the images tend to be of low magnification because of the sparse deployment of cameras in the farmland. Here, we present a 4.5× local zoom system for pest images of local high-magnification in a wide field of view. Such a system has a local zoom imaging channel for pest fine recognition and a peripheral imaging channel for searching pests with the same image plane. High-magnification imaging is made possible with fewer cameras for agricultural pest detection and recognition using the local zoom system. The experimental set-up is built to validate the system’s basic principle and is well used for the imaging of aphids on plant leaves. The results demonstrate that the system performs well for imaging of pests at different local magnifications.