pp-tinypose使用步骤

PP-TinyPose是PaddleDetecion针对移动端设备优化的实时关键点检测模型，可流畅地在移动端设备上执行多人姿态估计任务。

简单来说，给PP-TinyPose输入一张图片，PP-TinyPose能够告诉你，这张图片上的人的鼻子，肩膀，腕部等的坐标点。

一、下载tinypose

根据系统和python版本，在https://www.paddlepaddle.org.cn/whl/fastdeploy.html下载。

例如windows系统，python3.8则下载fastdeploy_python-1.0.7-cp38-cp38-win_amd64.whl。

二、使用命令安装

在安装文件目录下载执行pip install fastdeploy_python-1.0.7-cp38-cp38-win_amd64.whl安装。

三、下载模型文件

https://bj.bcebos.com/paddlehub/fastdeploy/PP_TinyPose_128x96_infer.tgz

解压文件，得到如下文件：

新建python文件，代码如下：

import fastdeploy
import cv2

model = fastdeploy.vision.keypointdetection.PPTinyPose('PP_TinyPose_128x96_infer/model.pdmodel', 
                                                       'PP_TinyPose_128x96_infer/model.pdiparams', 
                                                       'PP_TinyPose_128x96_infer/infer_cfg.yml')
img = cv2.imread('test.jpg')
result = model.predict(img)
print(result)

注意加载模型文件的目录。

结果说明：

KeyPointDetectionResult: [x, y, conf]
521.534119,-6.175730, 0.450121
529.790283,-26.832764, 0.342041
494.827026,-37.195068, 0.354224
533.464294,-34.687134, 0.379030
412.910248,-40.569958, 0.380281
626.002014,112.851463, 0.767896
293.472046,111.817810, 0.744751
726.036926,355.346588, 0.718352
184.570358,374.397095, 0.752293
737.261475,560.474304, 0.801744
433.852875,378.519501, 0.889975
632.677612,631.344238, 0.559384
445.482086,630.931213, 0.558599
667.912292,1020.837585, 0.637880
485.476807,1034.663818, 0.597461
681.528625,1173.857544, 0.062683
460.826141,1074.573120, 0.053947
num_joints:17

从上到下依次是

0: "Nose",

1: "Left Eye",

2: "Right Eye",

3: "Left Ear",

4: "Right Ear",

5: "Left Shoulder,

6: "Right Shoulder",

7: "Left Elbow",

8: "Right Elbow",

9: "Left Wrist",

10: "Right Wrist",

11: "Left Hip",

12: "Right Hip",

13: "Left Knee",

14: "Right Knee",

15: "Left Ankle",

16: "Right Ankle"

的图像坐标。