Dataset of Seven Fruit and One Vegetable Classes Captured for Agricultural AI Applications
Bursa 7th International Conference on Engineering and Mathematics, Bursa, Türkiye, 3 - 05 Temmuz 2026, ss.1-10, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Basıldığı Şehir: Bursa
- Basıldığı Ülke: Türkiye
- Sayfa Sayıları: ss.1-10
- Erciyes Üniversitesi Adresli: Evet
Özet
In agricultural production and the food supply chain, the automatic recognition, classification, and quality assessment of fruits and vegetables is of great importance. In this study, an image dataset that was created to enable the development and testing of agricultural artificial intelligence applications and that covers a total of eight produce classes, namely seven fruits and one vegetable, is presented. The dataset consists of a total of 8597 images belonging to the fruits pear, strawberry, apple, kiwi, lemon, banana, and orange, and the vegetable carrot. While the dataset was being created, a balanced distribution among the classes was ensured. Each image contains only a single product. In order to create a realistic dataset, rotten and defective samples, products at different color and ripeness levels, and partial images in which only a part of the product is visible were also deliberately added to the dataset. This diversity supports the developed models in working stably under non-ideal imaging conditions as well. The presented dataset can be used in the training and evaluation of deep learning-based tasks such as product classification, defect detection, ripeness analysis, and automatic product sorting. Thanks to its balanced structure and the realistic variations it contains, the dataset is a valuable resource for both academic research and precision agriculture and food processing applications.