English | [简体中文](README_CN.md) # PP-PicoDet + PP-TinyPose (Pipeline) Python Deployment Example Before deployment, two steps require confirmation - 1. Software and hardware should meet the requirements. Please refer to [FastDeploy Environment Requirements](../../../../../docs/en/build_and_install/download_prebuilt_libraries.md) - 2. Download the precompiled deployment library and samples code according to your development environment. Refer to [FastDeploy Precompiled Library](../../../../../docs/en/build_and_install/download_prebuilt_libraries.md) This directory provides the `Multi-person keypoint detection in a single image` example that `det_keypoint_unite_infer.py` fast finishes the deployment of multi-person detection model PP-PicoDet + PP-TinyPose on CPU/GPU and GPU accelerated by TensorRT. The script is as follows >> **Attention**: For standalone deployment of PP-TinyPose single model, refer to [PP-TinyPose Single Model](../../tiny_pose//python/README.md) ```bash # Download the deployment example code git clone https://github.com/PaddlePaddle/FastDeploy.git cd FastDeploy/examples/vision/keypointdetection/det_keypoint_unite/python # Download PP-TinyPose model files and test images wget https://bj.bcebos.com/paddlehub/fastdeploy/PP_TinyPose_256x192_infer.tgz tar -xvf PP_TinyPose_256x192_infer.tgz wget https://bj.bcebos.com/paddlehub/fastdeploy/PP_PicoDet_V2_S_Pedestrian_320x320_infer.tgz tar -xvf PP_PicoDet_V2_S_Pedestrian_320x320_infer.tgz wget https://bj.bcebos.com/paddlehub/fastdeploy/000000018491.jpg # CPU inference python det_keypoint_unite_infer.py --tinypose_model_dir PP_TinyPose_256x192_infer --det_model_dir PP_PicoDet_V2_S_Pedestrian_320x320_infer --image 000000018491.jpg --device cpu # GPU inference python det_keypoint_unite_infer.py --tinypose_model_dir PP_TinyPose_256x192_infer --det_model_dir PP_PicoDet_V2_S_Pedestrian_320x320_infer --image 000000018491.jpg --device gpu # TensorRT inference on GPU (Attention: It is somewhat time-consuming for the operation of model serialization when running TensorRT inference for the first time. Please be patient.) python det_keypoint_unite_infer.py --tinypose_model_dir PP_TinyPose_256x192_infer --det_model_dir PP_PicoDet_V2_S_Pedestrian_320x320_infer --image 000000018491.jpg --device gpu --use_trt True # kunlunxin XPU inference python det_keypoint_unite_infer.py --tinypose_model_dir PP_TinyPose_256x192_infer --det_model_dir PP_PicoDet_V2_S_Pedestrian_320x320_infer --image 000000018491.jpg --device kunlunxin ``` The visualized result after running is as follows
## PPTinyPosePipeline Python Interface ```python fd.pipeline.PPTinyPose(det_model=None, pptinypose_model=None) ``` PPTinyPosePipeline model loading and initialization, among which the det_model is the detection model initialized by `fd.vision.detection.PicoDet`[Refer to Detection Document](../../../detection/paddledetection/python/) and pptinypose_model is the detection model initialized by `fd.vision.keypointdetection.PPTinyPose`[Refer to PP-TinyPose Document](../../tiny_pose/python/) **Parameter** > * **det_model**(str): Initialized detection model > * **pptinypose_model**(str): Initialized PP-TinyPose model ### predict function > ```python > PPTinyPosePipeline.predict(input_image) > ``` > > Model prediction interface. Input images and output keypoint detection results. > > **Parameter** > > > * **input_image**(np.ndarray): Input data in HWC or BGR format > **Return** > > > Return `fastdeploy.vision.KeyPointDetectionResult` structure. Refer to [Vision Model Prediction Results](../../../../../docs/api/vision_results/) for the description of the structure. ### Class Member Property #### Post-processing Parameter > > * **detection_model_score_threshold**(bool): Score threshold of the Detectin model for filtering detection boxes before entering the PP-TinyPose model ## Other Documents - [Pipeline Model Description](..) - [Pipeline C++ Deployment](../cpp) - [Model Prediction Results](../../../../../docs/api/vision_results/) - [How to switch the model inference backend engine](../../../../../docs/en/faq/how_to_change_backend.md)