Departmental Bulletin Paper 植物病害自動診断に向けた葉領域検出および病変認識手法の提案

川崎, 雄介

Accurate plant disease diagnosis requires experts' knowledge but is usually expensive and time-consuming.So, an easy and accurate method is highly demanded in Japanese agricultural fields. In this study, we propose some ideas of the integrated system for detecting leaf and diagnosing its disease type. Our system achieved 84.0 % accuracy at the classification part under the 4-fold cross validation strategy.Key Words : plant disease, integral channel features, convolutional neural networks

Number of accesses :  

Other information