RKNNAI/RK182X-CNN-ResNet50V2

ResNet50V2

English | 简体中文

1. Model Introduction

ResNet50V2 is a 50-layer residual neural network for ImageNet image classification with 1,000 output classes. This repository provides RKNN deployment artifacts.

Field	Value
Model ID	RKNNAI/RK182X-CNN-ResNet50V2
Display Name	RK182X-CNN-ResNet50V2
Source Model	https://github.com/onnx/models/tree/main/validated/vision/classification/resnet
Model Type	CNN

Available Models

Configuration	Supported Chips	RKNN Runtime Version	Quantization	NPU Cores	Resolution
ResNet50V2-224x224-w8a8-1	RK1820, RK1828	-	w8a8	1	224x224

2. Files

Path	Description
README.md	English documentation
README_CN.md	Chinese documentation
LICENSE	License
ResNet50V2-224x224-w8a8-1/	Deployment configuration

3. Download

ModelScope

Full repository:

modelscope download --model RKNNAI/RK182X-CNN-ResNet50V2 --revision v1.2.0 --local_dir ./RK182X-CNN-ResNet50V2

Selected configuration and root documents:

from modelscope import snapshot_download

snapshot_download(
    "RKNNAI/RK182X-CNN-ResNet50V2",
    revision="v1.2.0",
    allow_patterns=["README.md", "README_CN.md", "LICENSE", "ResNet50V2-224x224-w8a8-1/**"],
    local_dir="./RK182X-CNN-ResNet50V2",
)

Hugging Face

Full repository:

hf download RKNNAI/RK182X-CNN-ResNet50V2 --revision v1.2.0 --local-dir ./RK182X-CNN-ResNet50V2

Selected configuration and root documents:

hf download RKNNAI/RK182X-CNN-ResNet50V2 --revision v1.2.0 --local-dir ./RK182X-CNN-ResNet50V2 --include README.md README_CN.md LICENSE 'ResNet50V2-224x224-w8a8-1/**'

4. SHA-256 Verification

From the directory where you ran the download command, enter the configuration directory and verify:

cd ./RK182X-CNN-ResNet50V2/ResNet50V2-224x224-w8a8-1
sha256sum -c SHA256SUMS

All entries must report OK before deployment.

5. Compatibility and Limitations

• Supported chips are listed per configuration above.
• Use the matching files from the same configuration.

6. Copyright and License

The upstream ONNX ResNet model is distributed under Apache License 2.0. See LICENSE and the upstream model documentation. These deployment artifacts were converted from resnet50-v2-7.onnx and quantized to W8A8.