AI Crop Disease Detection Using Mobile Camera

Abstract

Abstract-The paddy farming industry suffers a lot due to a number of diseases that are capable of producing a crop yield of 20-70 per cent. Conventional disease surveillance systems are slow, costly and need a person who is skilled and hence not accessible to small-scale farmers. The proposed AI-based system in this paper will involve the detection of the paddy disease through the remote sensing data of the Bhuvan and Bhoomi systems of the Indian Space Research Organization and the mobile camera. The given system uses a deep convolutional neural network (CNN) model that is mobile-oriented with an accuracy of 96.8 percent to recognize the major paddy diseases such as bacterial leaf blight, blast disease, brown spot, and sheath blight. The system is based on MobileNetV2 structure to perform efficient on-device inference with a mean processing time of 85ms per image. Combination with satellite Bhuvan imagery and Bhoomi land records allows monitoring of disease on a multi- scale; focusing on individual plants down to the area level. The process of field validation on 250 farmers confirmed the user satisfaction and the high rate of early disease detection increased considerably. The suggested system is a viable, economical, and accessible system of precision agriculture and food security. Index TermsPaddy disease detection, Mobile AI, Deep learn- ing, CNN, MobileNetV2, Remote sensing, Bhuvan, Bhoomi, Precision agriculture. Keywords: Paddy Disease Detection, Mobile AI, Deep Learning, MobileNetV2, Precision Agriculture.

Plant phenotyping relevance

モバイルカメラ画像とCNNによりイネの病害状態を直接推定する手法を開発し、精度・処理時間・圃場検証を報告しており、植物フェノタイピング手法が中心である。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.