VIGIA

PERFORMANCE

Benchmarking built for real-world inference.

We benchmark every detection model under identical conditions— measuring latency, throughput, and memory to surface trade-offs that matter in production.

Simple YOLO

Median latency (p50)

110.5 ms

Throughput

30.6 img/s

Peak memory

77.5 MB

ARGUS v8x

Median latency (p50)

109.3 ms

Throughput

27.6 img/s

Peak memory

88.6 MB

UltraFace

Median latency (p50)

3.3 ms

Throughput

1052.8 img/s

Peak memory

77.1 MB

import WhyThisModel from "@/components/WhyThisModel";

Detailed benchmark runs

Model	Batch	p50	p90	Throughput	Memory
Simple YOLO	1	148.8 ms	277.1 ms	5.1	111.3 MB
Simple YOLO	2	89.9 ms	161.8 ms	17.5	59.4 MB
Simple YOLO	4	102.9 ms	159.4 ms	33.8	72.9 MB
Simple YOLO	8	100.5 ms	190.3 ms	66.1	66.4 MB
ARGUS v8x	1	91.1 ms	189.5 ms	8.6	122.1 MB
ARGUS v8x	2	139.7 ms	201.0 ms	12.7	73.2 MB
ARGUS v8x	4	87.3 ms	154.8 ms	35.9	79.5 MB
ARGUS v8x	8	119.4 ms	243.8 ms	53.5	79.4 MB
UltraFace	1	3.3 ms	3.7 ms	293.1	77.0 MB
UltraFace	2	3.2 ms	3.5 ms	616.5	77.1 MB
UltraFace	4	3.2 ms	3.6 ms	1210.1	77.1 MB
UltraFace	8	3.5 ms	5.1 ms	2091.3	77.1 MB