# rknn-inspect
CLI tool for inspecting RKNN models.
> `rknn-inspect` is a command-line utility for inspecting and profiling Rockchip RKNN models. It helps developers understand model structure, input/output shapes and formats, SDK version info, and performance metrics. Useful during model development, conversion, and deployment on Rockchip NPUs.
## Usage
```shell
rknn-inspect --help
Usage: rknn-inspect [OPTIONS] <MODEL_PATH>
Arguments:
<MODEL_PATH> Path to the model file
Options:
--lib-index <LIB_INDEX> Which library path to use [default: 0]
-i, --io Show inputs and outputs
-n, --native-io Show native input/output information
--native-nhwc-io Show native input/output information in NHWC format
--native-nc1hwc2-io Show native input/output information in NC1HWC2 format
-p, --perf Enable performance profiling
-s, --sdk Show SDK information
--npu-cores <NPU_CORES> Select NPU cores to use [default: auto] [possible values: core0, core1, core2, core-all, core0-1, core0-1-2, auto]
--markdown Output in Markdown format
--full-name Show full name of the op in the --perf output
-h, --help Print help
```
## Installation
```bash
pipx install rknn-inspect
```
## Extended Usage
### View the input/output names, shapes, data types, and formats
```sh
rknn-inspect --io model.rknn
```
### View Performance Metrics
```sh
rknn-inspect --perf model.rknn
```
### View Performance Metrics for Specific NPU Core/s
```sh
rknn-inspect --npu-cores core0 --perf model.rknn
```
Raw data
{
"_id": null,
"home_page": null,
"name": "rknn-inspect",
"maintainer": null,
"docs_url": null,
"requires_python": null,
"maintainer_email": "Boyd Johnson <boyd@boundarybit.io>",
"keywords": "rknn, machine learning, rk3588, npu, deep learning, rknn-inspect",
"author": null,
"author_email": "Boyd Johnson <boyd@boundarybit.io>",
"download_url": "https://files.pythonhosted.org/packages/1f/6c/c6e153b3cead78f12abd124ac377b001c9fad12d2020edf37ea2f64631d6/rknn_inspect-0.2.0.tar.gz",
"platform": null,
"description": "# rknn-inspect\n\nCLI tool for inspecting RKNN models.\n\n> `rknn-inspect` is a command-line utility for inspecting and profiling Rockchip RKNN models. It helps developers understand model structure, input/output shapes and formats, SDK version info, and performance metrics. Useful during model development, conversion, and deployment on Rockchip NPUs.\n\n## Usage\n\n```shell\nrknn-inspect --help\nUsage: rknn-inspect [OPTIONS] <MODEL_PATH>\n\nArguments:\n <MODEL_PATH> Path to the model file\n\nOptions:\n --lib-index <LIB_INDEX> Which library path to use [default: 0]\n -i, --io Show inputs and outputs\n -n, --native-io Show native input/output information\n --native-nhwc-io Show native input/output information in NHWC format\n --native-nc1hwc2-io Show native input/output information in NC1HWC2 format\n -p, --perf Enable performance profiling\n -s, --sdk Show SDK information\n --npu-cores <NPU_CORES> Select NPU cores to use [default: auto] [possible values: core0, core1, core2, core-all, core0-1, core0-1-2, auto]\n --markdown Output in Markdown format\n --full-name Show full name of the op in the --perf output\n -h, --help Print help\n```\n\n## Installation\n```bash\npipx install rknn-inspect\n```\n\n## Extended Usage\n\n### View the input/output names, shapes, data types, and formats\n```sh\nrknn-inspect --io model.rknn\n```\n\n### View Performance Metrics\n```sh\nrknn-inspect --perf model.rknn\n```\n\n### View Performance Metrics for Specific NPU Core/s\n```sh\nrknn-inspect --npu-cores core0 --perf model.rknn\n```\n\n",
"bugtrack_url": null,
"license": null,
"summary": "A tool for inspecting RKNN models",
"version": "0.2.0",
"project_urls": {
"Homepage": "https://github.com/boundarybitlabs/",
"Repository": "https://github.com/boundarybitlabs/rknn-inspect"
},
"split_keywords": [
"rknn",
" machine learning",
" rk3588",
" npu",
" deep learning",
" rknn-inspect"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "3648e2c2a2c366861ac26f54d8ab4425b4f7229216ba6c8053d1df7fa65c8a40",
"md5": "3397609974eb9efff3721a121453bd7b",
"sha256": "2c77416416884f51e5db9c53fcf48bcb9a9ccbae4547428cbdaed993d6922543"
},
"downloads": -1,
"filename": "rknn_inspect-0.2.0-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl",
"has_sig": false,
"md5_digest": "3397609974eb9efff3721a121453bd7b",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 557184,
"upload_time": "2025-08-07T01:23:26",
"upload_time_iso_8601": "2025-08-07T01:23:26.331773Z",
"url": "https://files.pythonhosted.org/packages/36/48/e2c2a2c366861ac26f54d8ab4425b4f7229216ba6c8053d1df7fa65c8a40/rknn_inspect-0.2.0-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "62334a92410569e891ef00239b6690add3648fdde595256c36b993deae44eb91",
"md5": "a6bd343e111b690b23ad738a5192deff",
"sha256": "d8af8e68bd4f6c1e8854b86f8fe4d5961e0bf0d00116b9450ff219cd232cfbd0"
},
"downloads": -1,
"filename": "rknn_inspect-0.2.0-py3-none-musllinux_1_2_aarch64.whl",
"has_sig": false,
"md5_digest": "a6bd343e111b690b23ad738a5192deff",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 604757,
"upload_time": "2025-08-07T01:23:27",
"upload_time_iso_8601": "2025-08-07T01:23:27.542887Z",
"url": "https://files.pythonhosted.org/packages/62/33/4a92410569e891ef00239b6690add3648fdde595256c36b993deae44eb91/rknn_inspect-0.2.0-py3-none-musllinux_1_2_aarch64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "1f6cc6e153b3cead78f12abd124ac377b001c9fad12d2020edf37ea2f64631d6",
"md5": "65cf393fc2374a1cd2cdb1aa1889c463",
"sha256": "2df584f9bd397676f464b1a387fdba2231fcad4411b4526bcd443c7123243944"
},
"downloads": -1,
"filename": "rknn_inspect-0.2.0.tar.gz",
"has_sig": false,
"md5_digest": "65cf393fc2374a1cd2cdb1aa1889c463",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 12266,
"upload_time": "2025-08-07T01:23:28",
"upload_time_iso_8601": "2025-08-07T01:23:28.922575Z",
"url": "https://files.pythonhosted.org/packages/1f/6c/c6e153b3cead78f12abd124ac377b001c9fad12d2020edf37ea2f64631d6/rknn_inspect-0.2.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-08-07 01:23:28",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "boundarybitlabs",
"github_project": "rknn-inspect",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "rknn-inspect"
}
JSON
