The process:
We defined which parts of the system can be generic and camera-specific and need to be adapted to each case.
The pilot project includes retraining networks that are camera-specific, representing what networks see via data that is easy to manipulate. Then, we use this data with post-processing to meet our client’s specific needs.
SmartCat uses the database for training multiple models designed to extract different kinds of information from videos.
We perform an automatic secondary analysis of the collected raw data, so it can be used to create human-readable insights, alerting, and visualizations. As a result, Analytics about how many persons enter the space, how long they stay in each position, and which resources they use serve as business intelligence for clients to make smart decisions
Results
Optimization and utilization of the space
90% of people successfully detected and tracked
Automatization of manual operations
SmartCat collects relevant data from videos with high accuracy. Up to 90% of people and objects were successfully detected. This data is then automatically manipulated and transformed to accommodate specific client’s needs.
SmartCat improves the systems that support the algorithms applicable to security cameras
Smart Tip
Most CV algorithms hardly apply or apply with many difficulties to the security cameras, but we improve the Smart AI-powered systems that support those algorithms, using custom algorithms and optimization.