基于深度学习的石榴缺陷检测

Pomegranate Defect Detection Based on Deep Learning

摘要: 由于石榴的缺陷严重影响了其品质，文章基于深度学习方法对石榴缺陷进行识别与检测。根据国家林业局发布的“LY/T2135—2013石榴质量等级”行业标准，将缺陷分为鸟啄、裂果、腐烂、锈斑和日灼五种类型。采用YOLOv5模型和Mask R-CNN模型进行缺陷识别，检测精度分别为93.6%和88%。

Abstract: Because the defect of pomegranate seriously affects its quality, this paper uses deep learning method to identify and detect the defect of pomegranate. According to the industry standard “LY/T2135—2013 Pomegranate Quality Grade” issued by the State Forestry Administration, the defects are divided into five types: bird damage, fruit cracking, rot, rust and sunburn. YOLOv5 model and Mask R-CNN model were used for defect identification, and the detection accuracy was 93.6% and 88%, respectively.