2023年11月25日

Python代写｜Human Pose Estimation : Simplified

这是一篇美国的python代写作业案例分享

 

There is a dataset of human poses while exercising in the gym. The activities are named as the first column.

The pose-output.csv file is the result of feeding many exercise videos file into a pose estimation algorithm (i.e., Blazepose https://ai.googleblog.com/2020/08/on-device-real-time-body-posetracking.html).

Each exercise is a tensor (three-dimension data, i.e., x,y,z and time)

(1)  You need to create a visualization that as an input we feed an exercise and creates a visualization of the physical activity as an animation or gif. Look at 4th

Example gif: https://forknwork.wordpress.com/2018/02/14/openpose3d-pose-baseline/

You might need to rotate the data, to have a reasonable visualization.

(2)  For each exercise, identify joints that are moving, and list them. Besides, identify joints that are not moving as well.

(3)  List the degree of changes between joints. For example, take a look at following the output of your algorithm should be as follows:

{Moving: right_wrist, right_shoulder,Angle_changes: right_elbow {degree: 175, 170, … 30, 35, … 180} }

{Not_Moving: right_hip, right_knee, right_heel, left_shoulder, left_elbow, …}

(4)  (extra 5 credit) if you apply a clustering or classification algorithm on the result, that can distinguish different movements or group similar ones together you will get extra credit.