A Method for Insect Recognition in Passion Fruit Pulp Based on Improved YOLOv5

Effective detection and identification of insects on the passion fruit pulp production line is of great significance for product quality and food safety. This article is based on the YOLOv5 algorithm and uses deep learning to establish an algorithm model for identifying insects in passion fruit pulp. During the experiment, it was found that some images lacked model recognition accuracy due to factors such as poor lighting intensity and reflective lighting. This article introduces an adaptive histogram equalization algorithm to improve the model, enhance image contrast, and optimize the test image set by introducing two algorithms: low light image enhancement and reflection removal, to improve the model’s generalization ability and adaptability to different complex environments. The accuracy of model testing has been significantly improved. By training on a large dataset and optimizing images, this paper successfully achieved accurate identification of insects in the passion fruit pulp, providing a fast and efficient detection method for the industrial application of subtropical fruit processing.