A research team at CNU has developed an application and system that can identify foreign weeds through image recognition.

The joint research team of the CNU Intelligent Electronics Lab (Director: Professor Kim Jin-young) released the results of its research on November 26, 2021, in a joint ministry performance presentation of the 2021 Biodiversity-Threating Invasive Species Management Technology Development Project. The project was hosted by the Ministry of Environment, the Korea Environmental Industry and Technology Institute (KEITI), the Rural Development Administration, the Quarantine Headquarters, the Korea Forest Service, and the Korea Forestry Promotion Agency. 

The joint research team conducted a database construction project based on deep learning-based foreign weed identification system research for 4-year from 2018 to 2021, developed annotation tools for DB construction for deep learning, and developed a deep learning-based algorithm for the identification of foreign weed images. Finally, the research team developed an Android application and system for the image identification of foreign weeds.

In this system, a user can take a picture of weeds in the field with a smartphone camera, choose the targeted part in the picture, and find out the type of weeds by simply clicking the “Identify Foreign Weeds” button. The deep learning-based foreign weed image identification model identifies and shows the results. The server categorizes the input foreign weed image according to the determination result and also stores metadata such as the shooting date, time, and area information provided by the smartphone to build a database for the identification of foreign weeds in the future.

The research team consists of Dr. Kim Chang-seok (the National Institute of Food Science), Professor Lee Yong-ho (Hankyung University), Professor Kim Jung-gu (Chungnam National University), Kim Seung-cheol (Sungkyunkwan University), Professor Jeong Young-jae (Shin-gyeong University), and CEO Oh Young-joo (the Future Environmental Ecology Research Institute). The research team has conducted the research, entitled “Deep Learning-Based Foreign Weeds Identification System Development,” since 2018.