Deep Learning-Based Plant Disease Detection and Pesticide Recommendation System for Smart Agriculture

Abstract

Agriculture is an essential part of the worldwide economy, and initial detection of crop disease is essential to avoid substantial yield reduction. Conventional approaches of disease detection are primarily completed manually by specialists, which is expensive and frequently requires human errors. This survey aims to introduce an intelligent deep learning model to identify crop disease and suggest pesticides. This model is established on Convolutional Neural Networks (CNN) and uses the idea of transfer learning to sort the disease from the leaves of crops such as tomato and pomegranate. The proposed model works on the rule of image classification and is accomplished through image preprocessing, feature extraction, and classification employing the pre-trained model MobileNetV2. Once the disease is detected, it is mapped to the dataset.

Code and data availability

The paper describes a MobileNetV2-based crop disease detection system using leaf images, but provides no public URL, repository, or deposit for its dataset, code, trained model, or pesticide recommendation CSV. The dataset is only vaguely described as 'collected from publicly available plant disease image repositories'

No evidence-backed public reproduction asset is currently recorded.