A deep learning based approach for classifying the maturity of cashew apples.

Abstract

Over 95% of cashew apples are left to waste and rot on the ground. However, both cashew nuts and the often overlooked cashew apples possess significant nutritional and economic value. The cashew apple constitutes the major part (90%) of the cashew fruit, with the nut forming a modest portion (10%). Cashew nuts can be harvested and processed even after lying on the ground, but cashew apples are more delicate. Assessing the maturity status of these apples still requires human visual observation due to the challenges posed by their moisture content. Timely harvesting is crucial, as the pseudofruit is prone to microbial infections upon hitting the ground, making the process time- and labor-intensive. In this study, a Deep Learning based image classification model is presented, which can be used to automatically identify mature cashew apples. The model achieved an accuracy of 95.58% in classifying the cashew apples (immature vs. mature). Overall, the results highlight the potential of Deep Learning models for the classification of cashew apples and other fruits for precision agriculture purposes. This approach could enhance the harvesting process by enabling the utilization of the entire fruit and reducing the need for manual labor, thereby unlocking the full economic potential of the cashew tree.

Plant phenotyping relevance

カシューナッツ果実の成熟状態という植物器官の状態を画像から自動推定する深層学習手法が研究の中心であり、単なる収穫位置検出ではないため。

Code and data availability

The paper's Data Availability statement explicitly names the public cashew image dataset used for the maturity classification model, hosted on Mendeley Data (DOI 10.17632/r46c6bpfpf.1), with a related Data in Brief description article. No author analysis code or trained model checkpoint is reported.