Semantic Segmentation for Early Detection of Plant Stress and Disease using Multispectral and Thermal Imaging Through U-Net based Model

Abstract

Early detection of plant stress and sickness is more and more critical for sustainable agriculture, as diverse stress elements including drought, pests, nutrient deficiencies, and sicknesses-threaten plant fitness and productiveness. Traditional detection methods are often manual and time- in depth, main to delays in intervention. This paper presents a novel framework for early detection of plant stress and disease that combines U-Net based semantic segmentation with multispectral and thermal imaging. Using excessive-resolution multispectral and thermal imagery, U-Net can accurately section stress-affected regions in plant leaves and stems, distinguishing them from healthful tissue and taking into consideration well timed, focused intervention. And also this challenge is enhanced with disease prediction abilities, this version can also analyze environmental information, which include humidity, temperature, and soil great, to understand patterns that correlate with precise diseases. By integrating ancient data on preceding ailment outbreaks, U-Net's prediction layer identifies early signs and symptoms of recurring issues approximately the plant healthful or now not healthy by means of checking it is having any ailment or not. By reading stress patterns and distributions inside the segmented regions, the model provides a strong tool for early detection and unique intervention, probably decreasing yield losses and resource wastage in agriculture.

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.