Egocentric Retail Video Dataset
First-person video of real retail environments — grocery stores, pharmacies, department stores — with product interaction annotations for training retail automation AI.
Dataset at a Glance
Comparison with Public Datasets
How Claru's dataset compares to publicly available alternatives.
| Dataset | Clips | Hours | Modalities | Environments | Annotations |
|---|---|---|---|---|---|
| Metrabs Retail | 5K | 15 | RGB-D | Lab store | Pose, shelves |
| EgoProceL | 62 | 8 | RGB | Mixed | Procedure steps |
| Claru Retail | 70K+ | 500+ | RGB, Depth | 25+ store types | Products, shelves, paths, hands, navigation |
Use Cases
Shelf Monitoring Robots
Autonomous shelf scanning and out-of-stock detection using mobile platforms. Example models: Simbe Tally, Badger Technologies, BossaNova.
Shopping Assistant Robots
Customer interaction and in-store navigation assistance. Example models: Fellow Robots, SoftBank Pepper, LG CLOi.
Visual Retail Analytics
Understanding customer behavior and product interaction patterns. Example models: RetailNext, Trax, Standard AI.
Key References
- [1]Ragusa et al.. “The MECCANO Dataset: Understanding Human-Object Interactions from Egocentric Videos.” WACV 2021, 2020. Link
- [2]Grauman et al.. “Ego4D: Around the World in 3,000 Hours of Egocentric Video.” CVPR 2022, 2022. Link
- [3]Li et al.. “Ego-Exo4D: Understanding Skilled Human Activity from First- and Third-Person Perspectives.” CVPR 2024, 2024. Link
How Claru Delivers This Data
Claru collectors capture first-person video in real retail environments. Unlike ceiling-mounted CCTV datasets, Claru's egocentric perspective matches how shelf-scanning robots perceive their environment. Product-level interaction annotations enable training models that understand shopping behavior at a granular level.
Frequently Asked Questions
Grocery stores, pharmacies, convenience stores, department stores, home improvement stores, and specialty retail, ranging from 2,000 to 100,000+ square feet.
Every hand-product contact event is labeled with timestamps, product category, interaction type (pick up, examine, return), and shelf location.
Yes. Shelf-state annotations label visible products, positions, and gaps, training models to detect out-of-stock conditions and planogram deviations.
Request a Sample Pack
Get a curated sample of egocentric retail video data with full annotations to evaluate for your project.