노지 재배 고구마의 수분 스트레스 수준 평가를 위한 인공지능 기반 다중 영상 시스템

초록

Recent abnormal weather conditions in South Korea, including unexpected droughts and floods, have adversely affected the yield and quality of field-grown sweet potatoes. This has highlighted the critical need for an effective system to evaluate water stress in crops. In response, in this study, an artificial intelligence-based multi-imaging system was developed for assessing water stress in field-grown sweet potatoes. The system incorporates RGB image preprocessing, background removal, and machine learning models, specifically Convolutional Neural Network (CNN) and Support Vector Machine (SVM). The models achieved coefficients of determination of 0.80 for CNN and 0.86 for SVM. This system offers a reliable method for quantitatively evaluating water stress in sweet potatoes and managing irrigation. Furthermore, it holds potential for application in water stress assessment across various crops.

키워드