Crop Stress Detection Using AI

Abstract

Crops deal with all sorts of stress as they grow—things like missing nutrients, not enough water, or pests showing up where you don’t want them. If you catch these problems and you save the harvest and keep food production steady. But the old way of checking crops by hand? It’s slow, subjective, and honestly, just not practical for big fields. In this study, we built an automated crop stress detection system powered by AI. It uses image processing and deep learning, specifically a Convolutional Neural Network (CNN) based on the MobileNetV2 architecture. We set this up in TensorFlow and used the ImageDataGenerator function to keep our training data fresh and varied. The backend runs on Python, taking care of image preprocessing and making predictions. On the front end, we used ReactJS, so users can upload crop photos and instantly see what the system finds. The results speak for themselves. Our model hits 93.2% accuracy, with a precision of 91.5% and an F1-score of 92.3%. That’s solid proof this system works and can actually help farmers and researchers in real agricultural settings.

Code and data availability

The paper describes a MobileNetV2-based crop stress detection system but contains no public dataset, image collection, code repository, model checkpoint, or supplement with availability language. All referenced URLs are documentation (TensorFlow, Supabase, ReactJS) or cited prior work; no authors' public asset URL is给定

No evidence-backed public reproduction asset is currently recorded.