A study on paprika disease detection with YOLOv4 model using a customed pre-training method

Abstract

In this study, we employ object detection technique to detect 5 paprika diseases in greenhouse, namely Blossom end rot, Graymold, Powdery mildew, Spider mite, and Spotting disease. YOLOv4 model is used to detect the diseases in real-time with better detecting accuracy. Transfer learning is incorporated to enhance the detection performance where we employed 2 different methods to investigate the performance of the YOLOv4 backbone architecture. In first method, we only used the pre-trained YOLOv4 backbone on ImageNet classification dataset, whereas in second method, we tuned the pre-trained backbone weights (i.e. first method) with our own cropped paprika disease images. We found that the second method performed significantly better (results 4% higher mAP and reduces false positive considerably) than that of first method. Therefore, in this paper, we applied the second method for further evaluation of paprika diseases. Source code is at https://github.com/RoBoTics-JHJ/A-study-of-paprika-disease-detection-with-YOLOv4-model-using-a-customed-pretraining-method.