YOLO Papaya: Towards Identification of Maturity Stages of Papaya on Trees in Natural Environment

Papaya, a tropical fruit well known for its nutritional and medicinal value. The fruit I graded as per maturity – unripe and ripe is required for harvest operations. The yield estimation is also a required parameter for economic planning and market forecasting. Computer vision systems specifically object detection models in recent years have gained popularity in fruit detection. The proposed study addresses the critical need to automate the detection, classification, and counting of papaya fruits on trees. A self-collected diverse dataset, meticulously curated with papaya tree images representing various maturation stages, serves as the foundation for training algorithms to identify and classify ripe and unripe fruits amidst complex foliage. YOLOv5, YOLOv7 and YOLOv8 models are experimented with to detect the maturity stages of the fruit. F1, Precision, and Recall is determined for the 3 YOLO models. The YOLOv5 generated mAP of 0.87, 0.82 for YOLOv7 and 0.95 for YOLOv8 predicting the YOLOv8 performs better than other models.