Video Analytics when implemented properly are a force multiplier, when implemented incorrectly are a liability to the security solution.
Video Analytics at its core are intelligent processor based algorithms designed to detect and classify events. When implemented properly and functioning as intended with near 100% accuracy, Video Analytics are revolutionary force multipliers working tirelessly hour after hour. On the other hand if Video Analytics fail to accurately and consistently detect the event they are programmed to do monitor against, they cannot be trusted and become a liability to the security solution operations.
Two key metrics have emerged to rate Video Analytics:
- Probability of Detection (POD) - the reality of Video Analytics is that no system has a 100% POD in all situations. The very nature of video camera placement over the target area, the resulting lighting (Day and Night), angle of view, size of objects in the field of view, contrast, etc. make it impossible for every Video Analytic algorithm in every situation to have a perfect POD. Any company that claims 100% POD is only interested in the sale, not in the effectiveness your overall security solution.
- False Alarm Rate (FAR) - For the same reasons as above all Video Analytic algorithms will raise false alarms. It is impossible to have a FAR rate of zero (0). The liability of high FAR comes down to trust. Security system operators will quickly learn to dis-trust the Video Analytics and without thought will treat an increasing number of alarms as false alarms. The result will be that the operators will ignore a high percentage of alarm or more commonly disable the alarms creating a serious hole in the security solution.
- Implement a multi-layered security solution with Video Analytics being one layer of many where possible. Properly linking multiple technologies will raise the POD and lower the FAR increasing the robustness and reliability of the overall security solution.
- Implement scheduled testing and calibration of the Video Analytics to ensure the highest possible POD and lowest possible FAR over time.
The Video Analytics processor based algorithms information source is a 2-D video image. Most systems require the input of additional information such as camera height, tilt angle, horizon, camera image width or height, ground elevation changes, etc. enabling the algorithms to generate a mathematical 3-D environment making it possible to:
- The appearance of an objects
- The removal of an objects
- The movement of an object
- Textual Characters
- Distance from camera to object
- Object size
- Direction of travel of object
- Speed of travel of object
- Color analysis
Based on Video Analytic functionalities many different applications have been developed to detect and classify events. Every security solution has unique requirements, contact us for assistance in determining how or if Video Analytics are appropriate to your security solution. Some examples of how Video Analytics can be implemented are:
- Abnormal Behavior / Irregular Patterns
- Access Control – Video Verification
- Anti-Tailgating & Piggybacking
- Area Intrusion Detection
- Asset Protection
- Automatic PTZ Camera Tracking
- Bandwidth Management –for Remote Monitoring
- Camera Tamper Detection
- Counting (people, vehicles, etc.)
- Crowd Detection
- Customer & Time Analysis
- Face Detection and capture
- Graffiti detection
- Image enhancement
- Image stabilization
- License Plate/Number Plate Detection and capture
- Loitering Detection
- Motion Detection
- Object classification
- Object Left behind
- Object Tracking
- People running
- Perimeter Monitoring
- POS / Transaction Monitoring
- Queue monitoring
- Slip and Fall
- Vehicle Detection
- Vehicle Speed (Fast or Slow)
- Vehicle Traffic/Parking Violations
- Video Search
- Wrong Direction travel