Using computer vision to detect, classify, and map coffee fruits during harvest

Abstract

Computer vision is a technology that integrates image processing and pattern recognition to extract information from digital images. This research uses the you only look once (YOLO) method to detect, classify, and map the maturity stage of coffee fruit during harvest. The YOLOv3-tiny model was applied to classify coffee fruits into three categories including unripe, ripe, and overripe. Results showed an average precision of 86.0%, 85.2%, and 80.0%, respectively. The system enables more efficient harvest management by utilizing spatial and temporal information for coffee fruit quality mapping. This mapping can help farmers reduce operational costs and improve efficiency through the application of precision farming techniques. This technology proves great potential in improving the quality and quantity of coffee yields with a fast and accurate computer-based approach.