Pest diagnosis system based on deep learning using collective intelligence

With the recent development of agriculture, the growing area and utilization rate of facilities are increasing, but it is necessary to control and prevent pests, and if the disease is detected at an early stage, appropriate treatment is possible. To this end, researches on control systems using artificial intelligence are being expanded recently, therefore we propose a pest diagnosis system using data acquisition and deep learning through collective intelligence. This study modeled the diagnostic system based on deep learning using the collective intelligence that the user group participates in the prediction of pests arising from the plant cultivation and the data registered by experts in the field. Diagnostic data were collected information on pest diagnosis registered on the Internet and used; the collected data were constructed as a data set that is easy to analyze, through preprocessing, types of crops were classified, pests data were studied through TensorFlow. Most of the researches for the control and prevention of pests are based on web-based expert system. In this paper, we collect data through the collective intelligence and the general public. Especially, when a user uses input question and answers data without a formalized format, it gives wrong prediction; therefore, the preprocessing process was performed for data analysis because it could adversely affect the reliability of the system. After the data collection and preprocessing process is completed, a prediction model is created using TensorFlow, an artificial intelligence open source framework, using the generated data set. The user was allowed to input arbitrary data values while testing the data one to five times based on the data value and the effective value of the prediction model was confirmed according to the change of the value. Through the research, it is proved that diagnosis of pests is possible by using collective intelligence. In future research, research on the construction of a system that enables users to diagnose pests more easily should be continued using not only collective intelligence but also image and video.