ShatkoraSet: A Real-World Image Dataset for Freshness Classification of Shatkora (Citrus macroptera)

This dataset consists of images of shatkora (Citrus macroptera) fruits categorized based on their visual freshness condition. The data is organized into class-wise folders, each corresponding to a specific freshness category. The images were collected from real-world market environments and are intended for use in computer vision and machine learning tasks such as fruit freshness detection, quality assessment, and classification.
The dataset comprises 1,697 high-resolution RGB images of shatkora fruits captured in natural market settings in Lalbazar, Sylhet, Bangladesh. All images were acquired using a single smartphone camera under natural lighting conditions to maintain consistency while preserving real-world variability in illumination, background, orientation, and surface appearance. Each image has a resolution of 3000 × 4000 pixels and is stored in JPG format.
The images are organized into the following folders based on fruit condition:
Fresh: Images of shatkora fruits with intact surface texture, uniform color, and no visible signs of spoilage or decay (1,535 images).

Rotten: Images of shatkora fruits showing visible spoilage indicators such as discoloration, surface collapse, mold growth, and texture degradation (162 images).

Freshness labeling was performed based on observable visual cues commonly used in agricultural inspection and computer vision research. This dataset aims to support research in automated fruit quality assessment and address the lack of publicly available image datasets for region-specific fruits such as shatkora.