DeepLabv3+를 활용한 논콩 재배면적 분류 기술 개발

초록

This study aims to develop a deep learning-based crop classification to effectively assess the cultivation area expansion of paddy-grown soybeans in South Korea. As the cultivation area of paddy fields decreases and the cultivation of soybeans increases, accurately identifying the cultivation area of strategic crops has become crucial. To achieve this, we utilized high-resolution UAV imagery and the DeepLabv3+ deep learning algorithm. DeepLabv3+ can extract information from a wide receptive field, enabling high-accuracy classification and segmentation even in complex agricultural images. The results of this study indicate that the model can predict the cultivation area of soybeans with an accuracy of 94.86%, with a precision of 93.81% and a recall of 93.71%. This research is expected to provide valuable information for agricultural policy and crop management.

키워드