Oranges in the field

The orange on-field dataset is a large-scale, image collection under high environmental variability designed for fruit detection, classification, and maturity assessment in real-world orchard environments. It includes 5025 square images (pixel resolution : 640x640) captured under diverse weather, lighting conditions, and maturity stages. Images were collected using different smartphone cameras and preprocessed through a custom cropping algorithm to optimize annotation efficiency. The dataset was labeled using a semi-automated approach, while the human-in-the-loop was involved for verification and validation. The structured format and environmental diversity of the dataset make it valuable for AI applications in precision agriculture, such as fruit maturity assessment, yield estimation and automated harvesting.