Latency & Accuracy Tier Cheat Sheet

A one-page reference to match each surveillance analytic to the tier that meets its performance budget: the four-leg detection-latency budget (capture/encode, frame-to-compute, inference, result delivery) with edge ~25–120 ms vs cloud ~300–800 ms; the placement rule (under 200 ms = edge, 200 ms–1 s = hybrid, multi-second = cloud); the accuracy ceiling by tier (nano ~40% mAP to xlarge ~55%, quantization gap under 3–8%); the rule that accuracy is a precision/recall range (LPR 90–98%, face recognit