1.4 million grocery and pharmacy shelf photos from Israel and Europe, labelled with SKU-level bounding boxes, price tags and out-of-stock gaps for shelf analytics models.
Data product · Listing updated 11 Nov 2026
France, Germany, Italy, Poland
Israel
Box agreement above 0.85 IoU on audited samples
1.4M images, 31M bounding boxes, 85,000 SKUs
Startups, Small and medium businesses, Mid-market, Enterprises
Photos captured in partner stores with written permission of the store operators
Handheld and shelf-robot capture; boxes drawn by trained annotators with 10% double review
Real store shelves, photographed with the store operator's permission and labelled at product level.
Images are captured in partner stores in Israel, Germany, France, Poland and Italy under written agreements with store operators, using handheld phones and shelf-scanning robots. Shoppers and staff who appear in frame are blurred automatically and checked by a second annotator, and frames with clearly visible people are discarded. Boxes are drawn by trained annotators following a published guideline; 10% of images are double-labelled, and audited samples show box agreement above 0.85 IoU. Brand names that appear on packaging are captured as they appear on shelves. New stores and categories are added monthly.
12 fields as documented by Tavnit Vision Data.
| Field | Type | Description | Example |
|---|---|---|---|
| image_id | string | Image identifier | — |
| capture_date | date | Date the photo was taken | — |
| country_code | string | Country of the store | — |
| store_format | string | supermarket, convenience or pharmacy | — |
| camera_type | string | handheld_phone or shelf_robot | — |
| resolution | string | Image width and height in pixels | — |
| bbox_count | integer | Number of labelled boxes in the image | — |
| boxes | json | Box coordinates with class, SKU and brand labels | — |
| price_tags | json | Price tag regions with transcribed text | — |
| gap_regions | json | Empty facing and out-of-stock regions | — |
| occlusion_level | string | none, partial or heavy | — |
| people_blurred | boolean | True when any person in frame was blurred | — |
Real rows from the provider's sample file. Personal data is masked before anything is shown.
| image_idstring | country_codestring | store_formatstring | bbox_countstring | top_sku_labelstring | gap_regionsstring |
|---|---|---|---|---|---|
| img_100231 | IL | supermarket | 42 | Demo Cola 1.5L | 0 |
| img_100232 | PL | convenience | 27 | Example Crisps Paprika 150g | 2 |
| img_100233 | DE | pharmacy | 35 | Sample Sunscreen SPF 50 | 1 |
Prices in USD, as published by Tavnit Vision Data
per labelled image
50,000-image starter set
Converted amounts use DataStoa display rates and are approximate. Tavnit Vision Data quotes the final price, taxes and terms.
Request detailed pricing5 countries across 2 regions.
Time coverage: 2020 to present.
Provider-reported
Latency: Box agreement above 0.85 IoU on audited samples
DataStoa has not audited these figures. Ask the provider how they are measured, or test them on a sample.
The licence is agreed directly between you and Tavnit Vision Data. This summary is provider-reported; the linked terms prevail.
Not listed: internal analytics, AI retrieval and grounding, redistribution allowed, marketing and outreach. Ask Tavnit Vision Data if you need any of these.
Rights-cleared image, video and synthetic data for computer vision
Tavnit Vision Data captures and labels images and video for computer vision teams, working with retail and fleet partners under written agreements. It also generates synthetic datasets when real data is scarce or too sensitive to share.
Also available on: AWS Data Exchange