Honeycomb Cell Image Dataset for Seven-Class Classification of Honey Bee Comb Cells (7 Classes)

This dataset contains image fragments derived from high-resolution honeycomb images for the classification of honey bee comb cell states. The original honeycomb images were captured at a resolution of 2590 × 1940 pixels. The analysis region containing the honeycomb cells was cropped to 1162 × 574 pixels. The average width of a honeycomb cell within this region was determined to be approximately 14 pixels. Based on this measurement, each cropped image was divided into smaller image patches of 14 × 14 pixels. Each honeycomb image produced 3,403 image fragments arranged in 83 columns and 41 rows. The dataset was generated from 38 honeycomb images, resulting in a total of 129,314 image fragments. Each image fragment represents a single honeycomb cell and is labeled into seven different classes according to the content of the cell. These classes represent the different biological states of honeycomb cells and are defined as follows: 0 – Empty cell1 – Closed cell containing a larva2 – Open cell containing a larva3 – Closed cell containing honey4 – Open cell containing honey5 – Cell containing pollen6 – Unused cell The dataset was created to support research on automated honeycomb inspection, honey harvesting optimization, and computer vision applications in apiculture. The labeled dataset can be used for machine learning and deep learning models for image classification tasks.