Development of a Tool Using Computer Vision to Detect The Growth Stages of Fruits By Using Their Images of Premature Stage

This project presents the Development of a Tool Using Computer Vision to Detect the Growth Stages of Fruits by Using Their Images of Premature Stage. The system is designed to automatically classify fruit maturity levels—such as unripe, ripe, overripe, and rotten—using advanced deep learning techniques. Early and accurate detection of fruit ripeness is essential for improving post-harvest management, reducing food waste, and supporting automated sorting and grading in agricultural supply chains. To achieve high-precision recognition, the system integrates state-of-the-art YOLO (You Only Look Once) object detection models. Initially, YOLOv8 was evaluated due to its optimized architecture, improved feature extraction, and faster inference. However, for maximum accuracy and robustness, YOLOv9 was adopted as the final model. YOLOv9 provides advanced backbone design, enhanced multi-scale learning, and superior detection performance, making it well-suited for fine-grained classification of fruit stages even under varied lighting, occlusions, or background noise. A custom dataset containing multiple fruit types—such as apple, banana, mango, melon, orange, peach, and pear—was used for training and evaluation. The proposed tool preprocesses uploaded images, performs model inference, and displays annotated predictions alongside confidence scores. The system is deployed through a Flask-based web interface, ensuring fast processing, user accessibility, and a smooth workflow. Overall, this tool demonstrates efficient, accurate, and automated fruit maturity assessment, offering significant potential for agricultural automation, quality monitoring, and intelligent food-supply processes

Keywords:
Computer Vision, Fruit Growth Stages, Fruit Maturity Levels, Deep Learning, YOLO (You Only Look Once), YOLOv8, YOLOv9, Object Detection, Image Classification, Fruit Ripeness Detection, Post-harvest Management, Food Waste Reduction, Agricultural Automation, Automated Sorting and Grading, Multi-scale Learning, , Model Inference, Flask-based Web Interface, Fruit Types, Quality Monitoring, Food Supply Chain