Skip to content

Latest commit

 

History

History
35 lines (25 loc) · 1.68 KB

README.md

File metadata and controls

35 lines (25 loc) · 1.68 KB
Raw

English | 简体中文

PaddleDetection Deployment Examples for Python

Before deployment, the following step need to be confirmed:

This directory provides infer.py for a quick example of Picodet deployment on RKNPU. This can be done by running the following script.

# Download the deploying demo code.
git clone https://github.com/PaddlePaddle/FastDeploy.git
cd FastDeploy/examples/vision/detection/paddledetection/rknpu2/python

# Download images.
wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg

# copy model
cp -r ./picodet_s_416_coco_lcnet /path/to/FastDeploy/examples/vision/detection/rknpu2detection/paddledetection/python

# Inference.
python3 infer.py --model_file ./picodet_s_416_coco_lcnet/picodet_s_416_coco_lcnet_rk3568.rknn  \
                  --config_file ./picodet_s_416_coco_lcnet/infer_cfg.yml \
                  --image 000000014439.jpg

Notes

The input requirement for the model on RKNPU is to use NHWC format, and image normalization will be embedded into the model when converting the RKNN model, so we need to call DisableNormalizePermute(C++) or `disable_normalize_permute(Python) first when deploying with FastDeploy to disable normalization and data format conversion in the preprocessing stage.

Other Documents