Fruit Identification using YOLO v8 and Faster R-CNN –A Commparative Study

Fruits are rich in nutrients, vitamins, and minerals and play a crucial role in both dietary health and a nation's economy. In India, diverse fruit cultivation necessitates efficient identification methods. The manual fruit identification processes are time-consuming and biased, leading to agricultural inefficiencies. Hence, the adoption of machine learning and deep learning has gained immense popularity in the automation of the fruit identification preprocess. This study evaluates the performance of DL models, Faster R-CNN and YOLO v8, on a dataset featuring five fruits namely Apple, Cashew Apple, Banana, Mango and Orange for identifying fruits with the comparison between them. While both models yield impressive results, Faster R-CNN generally exhibits higher mAP. However, YOLO v8 excels in precision for specific fruits like Cashew Apple and Banana. The choice between models involves a trade-off between mAP and precision, emphasizing the need for further optimization considering factors like dataset variations, computational performance, and real-world applicability.