Knowledge graph and deep learning based pest detection and identification system for fruit quality
DingJu Zhu, LianZi Xie, BingXu Chen, Jianbin Tan, RenFeng Deng, Yongzhi Zheng, Qi Hu, Rashed Mustafa, Wanshan Chen, Shuai Yi, Kai Leung Yung, W.H. Ip
Abstract
Fruit usually plays a vital role in people's daily life. Many kinds of fruits are rich in vitamins and trace elements, which have high edible value. Pests and diseases are a considerable problem in the process of fruit planting. The quality and quantity of fruit can be effectively improved by the detection and preventing pests and diseases. However, suppose in the process of fruit growth, it is always necessary to manually identify and detect pests and diseases. In that case, it will inevitably consume a lot of workforce and material resources. Therefore, it is advisable to have an automated system to save unnecessary time and effort. This article introduces the detection and identification system of pests and diseases based on Raspberry Pi to identify and detect the pests and diseases of fruit such as Longan and lychee. Firstly, we constructed a knowledge graph of pests and diseases related to lychee and longan. Then, we used the Raspberry Pi to control the camera to capture the pests and diseases images. Next, the system processed and recognized the images captured by the camera. Finally, the Bluetooth speaker broadcasted the results in realtime. We constructed the knowledge graph through data collection, information extraction, knowledge fusion and storage. We trained the vgg-16 model, which achieves 94.9% accuracy in the pests identification task, and we deployed it on a Raspberry Pi.