Authors :Om Itankar, Sakshi Nirmal , Prajwal Meshram, Prof. Nilesh Korde
DOI : 10.46335/IJIES.2024.9.8.13
Abstract—: This paper focuses on the importance of image quality in fruit classification and certification processes, underscoring the significance of meticulous data collection in crafting robust learning models for this purpose. To address this necessity, we introduce the “FruitNet” dataset, which concentrates on six widely recognized Indian fruits. The FruitNet database comprises over 14,700 high-resolution images, categorized into three groups: "High Quality Fruit," "Low Quality Fruit," and "Mixed Quality Fruit." Each category includes images of apples, bananas, guavas, lemons, oranges, and pomegranates, all captured using a mobile phone equipped with a high-resolution camera. These images were intentionally captured against various backgrounds and lighting conditions to mirror real-world scenarios. This dataset serves as a valuable asset for training, testing, and refining fruit classification or recognition models. With its comprehensive collection of images, the FruitNet dataset facilitates the development of machine learning algorithms tailored to fruit analysis tasks.
