Knowing the customer’s concerns is critical to offering them the best experience for the systems integrator.
Facial recognition has become an integral part of physical security, but many customers are yet to take advantage of it. Several concerns still remain about their usage and efficiency, even though the benefits are quite clear.
According to Grandview Research, the global facial recognition market is expected to grow at a CAGR of 15.4 between 2021 and 2028. Security would account for a significant chunk of this growth.
However, this growth may not be without challenges. Some common complaints persist even among those who have already begun using facial recognition. This article explores such concerns with comments from three leading facial recognition solutions providers, especially in the access control sector.
Steve McMillen, Senior Director of Solutions Engineering for SAFR, RealNetworks
Problem: Users complain it takes too long to recognize their face
Solution: Ensure the face is well-lit and free of dark shadows in the eyes due to overhead lighting. If not possible, turn LED lighting on the access control device to add light onto the face.
Problem: The door is being unlocked by people in the background. People are authorized but are going to a different location and just happen to be in view of the camera.
Solution: Change the minimum face size threshold so that the system will only attempt to recognize faces at the terminal.
Problem: Terminal is triggering false spoofing alerts due to picture on the wall
Solution: Either add a crop region to ignore the face or if the face in the picture is small enough, change the minimum face size for detection to omit smaller faces in the background.
Kahren Arzoumanian, Co-founder and CTO, Kampr Systems
The biggest complaints are:
- Most facial recognition requires a server to support facial authentication readers.
- Most facial recognition in access control assigns the person the same ID as the person’s card ID. This reduces the security level to card ID and does not imply 100 percent a person as the card could be lost or borrowed. Facial recognition should be assigned as a separate user credential since many access control systems are not intuitive enough to distinguish the type of credential used for entry. A card has a much larger risk of theft than a biometric.
Dean Nicolls, Chief Marketing Officer, Oosto
- System integration
Integrating facial recognition with an existing camera and security infrastructure can be challenging. This requires that the vendor integrates with a variety of leading camera manufacturers, security door/turnstiles solution providers, and access control systems. It also requires system integrators who are familiar with these systems to help provide the connective tissue.
- Tuning the system
There is always a trade-off, where at extremely high true positive rates, false positives will also increase, and vice versa. If too many false positives and false negatives occur, the configurations of the system can be rebalanced by changing the threshold levels based on the specific use case. This is where tuning the system based on the customer needs to ensure the optimal balance is key.
- Camera placement
Integrators can help ensure optimal performance of the facial recognition system by where they place the cameras as well as the types of cameras being used.
- Edge computing
Integrators that are conversant with edge and near-edge compute devices can help lower the TCO of a video surveillance system with embedded facial recognition. This can amount to material cost savings.
Know the customer’s concerns
Knowing the customer’s concerns is critical to offering them the best experience for the systems integrator. As new solutions come in, issues will increase, especially when the technologies are complex. Period assessment of common concerns would help integrators respond to queries quickly.