Rockchip RK3588 NPU · yolov9-tiny

Frigate docs reference

Checked 2026-10-19· Frigate docs hardware page ·db-100

I run this too, add my numbers Opens the submit form prefilled with this hardware, model and version

Inference

35 ms

docs reference

Est. capacity

~29 det/s

1000 ÷ 35 ms · estimate

Cameras

—

not reported

CPU

—

not reported

Power

—

not reported

Parts

prices show only once verified by hand

Accelerator

Rockchip RK3588 NPU

RKNN (Rockchip) · 6 builds on DetectFPS · reported as “Rockchip RK3588 NPU (3 cores)”

price not tracked yet

Get price drops in the weekly newsletter

Price drops, once a week

When a tracked price is verified or drops, it goes in the newsletter alongside new benchmarks. No per-device alerts and no tracking: the newsletter is the only list.

Newsletter coming soon The signup form appears here as soon as the newsletter embed code is pasted into site.json. Issue #1 is already readable here.

Host

—

not reported

—

CPU

—

not reported

—

Cameras

—

resolution not reported

—

Tracked totalno verified prices yet

Setup

Detector

RKNN (Rockchip)

Detector instances

—

Model, as reported

yolov9-tiny

Model family · input

yolov9-t · input not stated

Camera resolution

—

Cameras

—

Frigate version

—

Numbers

dashes mean the source didn't say

Inference

35 ms

Inference range

—

Total fps

—

CPU

—

Power

—

Used in stats

Yes

Source

Where

docs.frigate.video/frigate/hardware

Date, as the source wrote it

—

Added to DetectFPS

copied from the Frigate project's hardware page on 2026-10-19; approximate reference value

What this build implies

~29 detections per second

1000 ÷ 35 ms. No camera count was reported, so use the calculator to see what that covers: at 5 fps and Typical activity, that is about 6 cameras comfortably.

Open in the calculator

Hardware

Rockchip RK3588 NPU

Detector type

RKNN (Rockchip)

Builds on DetectFPS

Median, this model

35 ms (1 build)

Price

price not tracked yet

Shown because this guide lists Rockchip RK3588 NPU in its hardware (hardware_ids). Cross-device guides are only linked when they name this device or its host.

Affiliate links

DetectFPS may earn a commission when you buy through links on this site. It never changes which builds we show or how they rank. Buy buttons say “affiliate link” whenever one is used; none is used on this page.

Similar builds

Same device, then the same model family

Rockchip RK3588 NPU

RKNN (Rockchip)

older version, may be outdated

Speed40 ms

Cameras1

Modelyolov8

Version0.12.1

AcceleratorRockchip RK3588 NPUprice not tracked
HostOrange Pi 5Bprice not tracked
CPU—

Sep 2023GitHub #7601

Rockchip RK3588 NPU

RKNN (Rockchip)

Frigate docs

Speed14 ms

Cameras—

Modelyolox-nano

Version—

AcceleratorRockchip RK3588 NPUprice not tracked
Host—
CPU—

checked 2026-10-19Frigate docs

Rockchip RK3588 NPU

RKNN (Rockchip)

Frigate docs

Speed18 ms

Cameras—

Modelyolox-tiny

Version—

AcceleratorRockchip RK3588 NPUprice not tracked
Host—
CPU—

checked 2026-10-19Frigate docs

Rockchip RK3588 NPU

RKNN (Rockchip)

Frigate docs

Speed20 ms

Cameras—

Modelyolo-nas-s

Version—

AcceleratorRockchip RK3588 NPUprice not tracked
Host—
CPU—

checked 2026-10-19Frigate docs

Rockchip RK3588 NPU

RKNN (Rockchip)

Frigate docs

Speed30 ms

Cameras—

Modelyolo-nas-m

Version—

AcceleratorRockchip RK3588 NPUprice not tracked
Host—
CPU—

checked 2026-10-19Frigate docs

Nvidia GTX 1080

ONNX (Nvidia / GPU) · Stated "28-30+ ms".

Speed28–30 ms

Cameras—

Modelyolov9-t 320

Version—

AcceleratorNvidia GTX 1080price not tracked
Host—
CPU—

Aug 25, 2025GitHub #19741

5 other builds use the same device; the rest share the yolov9-t model family.