Study
Supermarket chain · Europe
At a glance
- Client
- Supermarket chain · Europe
- Challenge
- Theft and suspicious gestures on the floor, plus opening, closing and aisle cleaning that were hard to check across the chain.
- Solution
- Agents designed to detect theft and suspicious gestures and to follow store opening, closing and aisle cleaning.
- Result
- 80% recall, 90.9% precision and 74% accuracy in detection, with alerts in 15 seconds on average.
- Recall
- 80%
- Precision
- 90.9%
- Accuracy
- 74%
- Average time to alert
- 15 s
01
Context
In a supermarket, theft blends into normal aisle traffic. Opening, closing and cleaning happen when no manager is on site, and checking them store by store does not scale.
02
How the agent was designed
The agents do two kinds of reading. One detects theft and suspicious gestures on the floor and alerts the store team. The others follow store opening and closing and aisle cleaning, comparing what happened with the planned routine.
03
What changed
For theft and suspicious-gesture detection, the agent reached 80% recall, 90.9% precision and 74% accuracy, and the alert reaches the team in 15 seconds on average. Opening, closing and cleaning are now checked through the reading, without site visits.
Next step
See what the reading finds in your chain.