# RKNN-Toolkit2

Repository: https://github.com/airockchip/rknn-toolkit2
Canonical: https://ross.abutalabs.com/products/rknn-toolkit2
Language: C
License: NOASSERTION
License Family: other
Last push: 2025-07-29T05:08:10+00:00

## Health v2 (maintenance only)
Score: 36/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 34, release rhythm 16, longevity 76
- inputs: {"age_days": 1073, "days_push": 400, "days_rel": 511, "gap_med": 101.0, "n_releases_24m": 3}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3313, forks 386 (observed 2026-08-28T04:07:56.049648+00:00)

## What it is
RKNN-Toolkit2 is Rockchip's SDK for converting trained neural network models into RKNN format and deploying them on Rockchip NPU chips like RK3588. It includes PC-side model conversion and evaluation tools plus Python and C/C++ runtime APIs for on-device inference.

## Use cases
- convert onnx or pytorch models to run on rockchip rk3588 npu
- deploy a yolov8 model on an embedded rockchip board
- run computer vision inference on rk3566 npu
- quantize a deep learning model for edge deployment
- benchmark model performance on rockchip npu
- run inference on rv1106 low-power vision chip

## When to choose
- you are deploying AI models to Rockchip SoCs with an NPU (RK3588, RK3576, RV1106, etc.)
- you need model conversion, quantization, and on-device inference in one toolchain
- you want C/C++ or Python APIs for NPU-accelerated inference on embedded Linux

## When to avoid
- you target non-Rockchip hardware such as NVIDIA GPUs, Jetson, or generic CPUs
- you are using older Rockchip chips like RK1808, RV1126, or RK3399Pro (use rknn-toolkit instead)
- you need to deploy large language models - use the separate RKNN-LLM SDK

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, llm-inference, sdk, compiler
- domain: machine-learning, embedded-systems, computer-vision, deep-learning
- platform: python, cpp, embedded, cli
- tags: rockchip, npu, model-conversion, rknn, edge-ai, inference, linux

## Member repositories
- airockchip/rknn-toolkit2 (main) score 36
- airockchip/rknn_model_zoo (examples) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:56.049648+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:42:21.957783+00:00, confidence not recorded.
  - readme: https://github.com/airockchip/rknn-toolkit2 (fetched 2026-08-28T04:07:56.049648+00:00, sha cbf671dfc07f)
- Data as of 2026-08-30T08:39:29.467469+00:00.