Human Pose Estimation and 3D posing on Unity: Step 1

Step 1: Human Pose Estimation

I mentioned about the Human pose estimations article on this “page” and I clone GitHub repo and everything work fine. For this source code, I create new anaconda environment because I used the different OpenCV version. Also, I didn’t use the Tensorflow GPU. You can use the GPU version it also creates much more estimated results.

Now, Pose estimation is working good but I need the send this key points on the Unity. First of all, I must the understand the code and I understand the which code ‘s results give the key points array. If I can create a connection with Unity also I can create the 3D skeleton with Vuforia for the Augmented Reality

When running the with single photo results like these;

In the code I saw codes I noticed the


System close when come to this line and bellow of this code’s purposes are 3D key-points modelling wih matplotlib.

When I delete sys.exit(0) 3D plot shows up

pose_3d = poseLifting.compute_3d(transformed_pose2d, weights)

pose_3d is the key points of the graph so that if I can send these values with internet to Unity Platform I can move the skeleton based on the human body moving. When I print pose_3d output like that;



6 thoughts on “Human Pose Estimation and 3D posing on Unity: Step 1

  1. Your posts about Human Pose Estimation and 3D posing on Unity are very helpful for me.But in the process of running, I have 2 problems.When I delete sys.exit(0),3D plot didn’t show.And how to print pose_3d output?Hope to reply me.

  2. Thanks for one’s marvelous posting! I actually enjoyed reading
    it, you could be a great author. I will make certain to bookmark
    your blog and definitely will come back someday.
    I want to encourage you to continue your great work, have
    a nice holiday weekend!

Leave a Reply

Your email address will not be published. Required fields are marked *