Citrus fruit detection and classification: Field-based images and FIDS-30 dataset

The aim of this study is to introduce a deep learning (DL)-based approach that uses a convolutional neural network (CNN) to automatically detect and classify citrus fruits, resulting in precise outcomes. The model was developed and tested using field images of citrus fruits from a remote place in southern India, as well as an extensive dataset of 971 images across 30 fruit types (FIDS-30). Pre-processing, segmentation, feature extraction, and fruit categorization are all part of the process, which is powered by CNN. The model's robustness was ensured by the research's use of effective data augmentation techniques such as rotation, flips, and color jittering, which expanded the dataset and enhanced its ability to handle diverse real-world scenarios, leading to better prediction accuracy. The model demonstrated exceptional performance with 97% precision, 98% accuracy, recall, and F1-score, effectively distinguishing citrus fruits from other types. It is advised to the community that effective data augmentation during the pre-training period is an essential step before applying advanced DL methods to data.