[Multi-person pose estimation] AlphaPose_yolov5 reasoning demo environment configuration

This part of the individual has already been reproduced. For some reasons, it cannot be open sourced. Roughly make a simple environmental record and a small summary.

The speed of using yolov5s is very fast. However, the detection accuracy may be worse than that of yolov3-spp. Due to the top-down multi-person human body pose estimation and tracking, detection problems will affect the final input sent for pose estimation, and ultimately affect its performance.

According to the analysis, we can basically know that Alphapose tracking uses target tracking and pose estimation.

1. Environment

torch==1.7.0
torchvision==0.8.1	
opencv-python==4.1.0.25
tqdm
natsort
Cython
pycocotools
scipy
terminaltables
wf-pycocotools==2.0.1.1
scikit-image>=0.16.2
googledrivedownloader
protobuf>=3.8.0
websocket-client

python setup.py build develop --user

apt-get update
apt-get -y install python3.6-tk