IMAGE CLASSIFICATION-BASED CROP YIELD PREDICTION FRAMEWORK FOR PRECISION AGRICULTURE DECISION SUPPORT

Abstract

This research focuses on improving crop yield prediction using classified satellite and drone imagery combined with deep learning techniques [1]. By using real-world datasets like PlantVillage and MODIS, the study enhances image-based classification for detecting crop type, plant health, and stress conditions. The proposed hybrid model uses convolutional neural networks to extract features and regression layers to correlate them with historical yield data. Preprocessing steps such as normalization and segmentation ensure higher model accuracy. The integrated framework achieves over 90% prediction accuracy, outperforming traditional models. Visual tools like heatmaps provide actionable insights for farmers to optimize inputs. This approach supports precision agriculture by enabling early decision-making based on crop condition. This model is scalable for various crops and different regions. The study highlights the potential of AI in transforming agriculture through smarter, data-driven strategies.