Skip to content
Karmyug

Retail product recognition

Supermarket shelves and individual products with tight bounding boxes.

N/ARetail & e-commerce

Overview

Images captured in real stores across mixed lighting with 500+ SKU classes. Roughly a third of the set is deliberately hard: occlusion, glare on packaging, partially empty shelves. That is where checkout and planogram models fail.

Specification

Resolution
1920x1080 and 4K
Format
JPG
Annotation
2D box, COCO and VOC

Sample

Illustrative sample. It shows the structure, fields and metadata of this dataset with invented values, not real recordings or records. A real sample pack is sent on request.

Sample records for Retail product recognition
filewidthheightdevicelightinglabel_typelabel_summaryannotator_idqa_statusconsent_ref
sample_0001.jpg19201080Android mid-range phonedaylightbounding boxes3 itemsann_0040acceptedconsent_03100
sample_0002.jpg30244032Android budget phoneindoorbounding boxes5 itemsann_0043acceptedconsent_03101
sample_0003.jpg10801920iPhonelow lightbounding boxes2 itemsann_0046accepted after reviewconsent_03102
sample_0004.jpg40323024Android mid-range phonedaylightbounding boxes7 itemsann_0049acceptedconsent_03103
sample_0005.jpg25601440Laptop microphoneindoorbounding boxes4 itemsann_0052acceptedconsent_03104

Quality assurance

  • Double-blind annotation
  • IoU above 0.9 against gold
  • Manual spot checks

Want a real sample before you commit?

The sample above shows the structure. We also send a representative slice that includes the hard cases, plus the QA report and licence terms. No call required.