Integrating YOLOv8 Architecture in a Mango Flowers Detection App Development

Keeping tabs on the vitality of mango blossoms is essential for improving farming techniques and guaranteeing a harvest. This study introduces a tool that utilizes the Yolov8 model to categorize mango blossoms as either vibrant or not. Users can submit pictures of mango blossoms via the application, which then follows a set of procedures including image adjustment, data enhancement, version 6 recognition, formation of bounding boxes determining confidence scores applying thresholds and non-maximum suppression (NMS). The application then tallies the quantity of non-vibrant flowers and computes the mean confidence levels before showcasing the overall detection accuracy. The system’s design incorporates a user interface that consists of a dashboard, for uploading images and viewing results; a control panel for adjusting detection settings; and a logging system to track previous detections. The backend integrates Yolov8 model for precise detection, in real-time processing. Evaluation criteria, like the confusion matrix and precision are employed to gauge how well the model performs in distinguishing between unhealthy mango flowers accurately based 0 n their level and other factors. The findings clearly show that the system can effectively differentiate between the two types of mango flowers and offer guidance, for farmers and agricultural supervisors. This research paper adds value to the precision agriculture sector by introducing a tool, for tracking the status of mango flowers to improve crop management and decision-making processes.