Test results

Performance testing

For performance testing, we took a video of a corridor with a flow of people moving at a speed of 5-10 people/sec. The volume of the database for identification is 249 persons.

Metrics

Metric name	Description
Maximum	Maximum value of parameters (CPU load, GPU load, amount of RAM, amount of VRAM) for the entire test time
95th percentile	CPU is loaded at no more than the specified value for 95% of the test time GPU is loaded at no more than the specified value for 95% of the test time The amount of RAM used does not exceed the specified value for 95% of the test time The amount of VRAM used does not exceed the specified value for 95% of the test time
Median	CPU is loaded at no more than the specified value for 50% of the test time GPU is loaded at no more than the specified value for 50% of the test time The amount of RAM used does not exceed the specified value for 50% of the test time The amount of VRAM used does not exceed the specified value for 50% of the test time
Recall	Percentage of identified faces
Precision	Percentage of correct identification
Max latency	Max time spent between event generation and submission
Total events out of 117	This number shows how many identification events were generated during the test (Maximally the test result should contain 117 identification events)

Test

Test machine specification

CPU	RAM	Graphics Card
AMD Ryzen 9 5950X @ 3.4 GHz: 16 cores (32 threads)	118GB DDR4	NVIDIA GeForce GTX 1080 Ti

Test results

	Number of video streams	1	5	8	10	12	15	18
Maximum	Number of CPU cores	1	6	12	14	16	20	24
	GPU load	16%	97%	93%	96%	96%	96%	89%
	Amount of RAM	3.83 GB	6.80 GB	9.27 GB	10.38 GB	12.36 GB	14.59 GB	17.30 GB
	Amount of VRAM	4.49 GB	4.72 GB	4.78 GB	5.23 GB	5.23 GB	5.30 GB	5.30 GB
95th percentile	Number of CPU cores	1	6	12	14	16	20	24
	GPU load	9%	61%	68%	60%	57%	54%	59%
	Amount of RAM	3.46 GB	6.80 GB	9.14 GB	11.12 GB	11.74 GB	13.72 GB	16.32 GB
	Amount of VRAM	4.49 GB	4.72 GB	4.72 GB	5.23 GB	5.23 GB	5.23 GB	5.30 GB
Median	Number of CPU cores	1	6	11	13	15	18	22
	GPU load	8%	44%	58%	47%	46%	36%	25%
	Amount of RAM	3.21 GB	6.42 GB	8.53 GB	9.64 GB	11.00 GB	12.73 GB	14.96 GB
	Amount of VRAM	4.49 GB	4.72 GB	4.72 GB	5.23 GB	5.23 GB	5.23 GB	5.30 GB
	Recall	99.10%	99.10%	98.30%	97.40%	96.60%	94.90%	92.30%
	Precision	100%	100%	100%	100%	100%	100%	100%
	Max latency (sec)	8	8	11	15	15	17	21
	Total events out of 117	116	116	115	114	113	111	108

Conclusion

~ 1.5 cores per video stream when using GPU in the most high-load case "Safe City".