Key facts
Contents & fields
The dataset contains 16,592 agri-food retail images divided into unlabeled and labeled subsets. The unlabeled subset is suitable for classification, object detection, and product recognition; the labeled subset includes 2,416 samples with detailed centroid annotations for on-shelf availability estimation, counting, or multi-task learning. Images were captured using an iPhone 14 Plus (1080p, 30fps, HDR) and an Intel RealSense Depth Camera D435i under varying shelf inclinations, lighting levels, and angles.
- Images——RGB images of agri-food retail scenes
- Centroid annotations——Center point coordinates of each object in the labeled subset
Research uses
Suitable for computer vision tasks in smart retail such as automated inventory monitoring, product recognition, on-shelf availability estimation, object detection, and real-time retail analytics.
Information comes from the source page. Please check that page for current details and terms.
Keywords
Why this is hard to get on your own
Obtaining a high-quality, multi-source, real-world retail image dataset for training and evaluating computer vision models.