A Diagnosis and Prescription System to Automatically Diagnose Pests

Crop losses continue to increase due to climate change and the presence of foreign pests. However, it is difficult for farmers to reduce crop losses because they cannot diagnose and prescribe against pests quickly enough. Therefore, in order to resolve the issue, this paper describes a mobile-based, automatic system for pest diagnosis and prescription using a smart device with which to diagnose pests and obtain prescription information by taking photographs. In order to diagnose pests, image searches based on similarity are conducted. Due to the features of the image-similarity search, sufficient data sets must be obtained in order to increase search precision. In order to increase the pest-image data set, images of pests were collected using a focused web crawler on the Internet. However, because there are many images that do not pertain to the applicable pests, the precision of diagnosis is reduced. Therefore, image precision was increased through an inspection system that utilizes experts. The images obtained with the pest-image collection and search system is indexed using an image similarity-based search system. Next, the similarity of images of pests taken with the user’s cell phone is compared, and information on diagnosis and prescription is shown to the user in real time. When the system-diagnosis performance was measured using three crops (pears, strawberries and grapes), a precision level of 83% was recorded. In the future, a video recognition-based system will be incorporated in order to enhance the precision and scope of application. It is expected that the system of mobile-based automatic pest diagnosis and prescription will provide quick, precise pest information to farmers and assist in the prompt prevention of disasters so as to minimize economic losses.