Enabling Facial Recognition for Smart Surveillance
Security Control | Identify Verification | Mask Detection | Temperature Measurement
People per hour per channel
(avg. appearing time per person = 3 secs)
(Total 19,890 people/hour)
(Total 41,310 people/hour)
2x Performance
(Total 77,520 people/hour)
3.9x Performance
People per hour per channel
(avg. appearing time per person = 3 secs)
(Total 17,550 people/hour)
(Total 39,000 people/hour)
2.2x Performance
(Total 74,100 people/hour)
4.2x Performance
People per hour per channel
(avg. appearing time per person = 3 secs)
(Total 19,890 people/hour)
(Total 41,310 people/hour)
2x Performance
(Total 77,520 people/hour)
3.9x Performance
People per hour per channel
(avg. appearing time per person = 3 secs)
(Total 17,550 people/hour)
(Total 39,000 people/hour)
2.2x Performance
(Total 74,100 people/hour)
4.2x Performance
* Video resolution:1080p, Frame rate 15FPS
* Facial recognition framerate: 12FPS
* FaceMe® Security 2.0.0 is used in the benchmark
* Tested Platforms: Xeon Silver 4214, NVIDIA RTX 2080 Ti, 128 GB Memory, Windows 10 Enterprise LTSC
People per hour per channel
(Total 500 people/hour)
(Total 1,000 people/hour)
People per hour per channel
(Total 500 people/hour)
(Total 1,000 people/hour)
* Video resolution: 1080p, Frame Rate: 10FPS
* FaceMe SECURITY 2.5.0 is used in the benchmark
* Tested on Windows 10 Pro (Core i3) and Win10 Enterprise, both with 16GB memory
* Camera resolutions set at 1080p, each camera running face recognition at 10 frames per second (FPS)
* Internal testing detected 1,000 faces per hour on each IP Camera video, with 2~3 faces per frame and 10 FPS for each camera.
