From AI Based Object Detection Model to Grape Yield Mapping for Precision Agriculture Applications

Yield mapping in vineyards is crucial for agronomic and economic management, allowing for precision operations like pruning, harvesting, fertilization, irrigation, and soil management. This leads to optimized resource use, improved grape quality, and increased productivity. Traditional yield mapping relies on expensive grape harvesters, which can cause grape loss and are not suitable for manually harvested vineyards. This research proposes a low-cost framework combining hardware and computer vision to generate yield variability maps, which can be used to create management zones for precision agriculture applications. The Yolov8 obtained the best performance with an overall precision of 89%. This approach aims to reduce reliance on costly machinery, enhance data accuracy, and make precision agriculture more accessible and effective.