A Survey on Deep Learning-Based Stress Detection and Classification in Arecanut Trees for Precision Crop Health Monitoring

Abstract

Arecanut cultivation is an economically important agricultural activity in many tropical regions; however, crop productivity is frequently threatened by a wide range of biotic and abiotic stress factors, including diseases, pest infestations, nutrient deficiencies, drought, and environmental fluctuations. Early identification of these stress conditions plays a significant role in mitigating crop yield losses and improving crop management practices. Recent progress in Machine Learning (ML), Artificial Intelligence (AI), and Deep Learning (DL) has enabled the development of automated systems capable of detecting plant stress with high accuracy using image-based and sensor-driven data. This paper presents a survey of recent studies employing Convolutional Neural Networks (CNNs), hyperspectral imaging, Vision Transformers (ViTs), UAV-based sensing, transfer learning models and multimodal learning approaches for stress detection and crop health monitoring. Existing methodologies are analyzed with respect to their performance, strengths, and limitations. Despite significant advancements, challenges such as limited annotated datasets, environmental variability, computational complexity, and inadequate real-field validation continue to hinder practical deployment. Emerging technologies including Explainable AI (XAI), Internet of Things (IoT)-based monitoring, edge computing, and precision agriculture offer promising opportunities for improving detection accuracy and scalability. The survey highlights current research trends, existing challenges, and future directions for developing reliable and efficient arecanut stress detection systems.