Predicting Calamansi Fruit Yield Using CNN-Based Flower Detection: A Deep Learning Approach

As agriculture integrates advanced technologies to enhance productivity, precise fruit yield estimation emerged as a need. Calamansi, a citrus fruit in Southeast Asia, demands accurate yield projections for effective resource allocation and market responsiveness. Traditional methods lack efficiency, driving the exploration of innovative techniques like Convolutional Neural Networks (CNNs), and You Only Look Once (YOLO). This paper introduces a deep learning approach for calamansi fruit and flower detection, supported by a customized dataset. YOLOv8m model was used for calamansi and flower detection, achieving an F1 score of 0.815, a Recall of 0.794, a Precision of 0.839, and mAP@0.5 of 0.875. The study concluded that YOLOv8m algorithm can significantly enhance the production of fruits in calamansi farms.