PoseCamera is python based SDK for human pose estimation through RGB webcam.

Hi,

Thank you for this lovely SDK. I am trying to explore this library. My intension is to calculate the movement of individuals using pose estimation.

How can I implement PoseCamera on video/webcam data in realtime? And get the estimation value in a small buffer data (maybe of 4 secs ) to further process it, to find the movement in each individual. I am trying to use google colab for it. Any example notebook?

Regards, Suti

Hi ,

Thank you for PoseCamera!

I installed it on my Linux machine and trying to run it on the Jupyter notebook. But I get at the below error:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
      4 # !wget https://storage.googleapis.com/wt_storage/lightweight_pose_estimation.pth
      5 
----> 6 posecamera.load("lightweight_pose_estimation.pth")
      7 
      8 image = cv2.imread("3.jpeg")

**AttributeError: module 'posecamera' has no attribute 'load'**

On Linux with nvidia-docker2

fabien@fabien-CORSAIR-ONE-i160:~$ xhost +; docker run --name posecamera --rm --gpus all -e DISPLAY=$DISPLAY --device=/dev/video0:/dev/video0 wondertree/posecamera --video=0
access control disabled, clients can connect from any host
qt.qpa.xcb: could not connect to display :1
qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in "/home/user/miniconda/lib/python3.8/site-packages/cv2/qt/plugins" even though it was found.
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: xcb.

Same for running on image rather than webcam :

fabien@fabien-CORSAIR-ONE-i160:~$ docker run --name posecamera --rm -e DISPLAY=$DISPLAY  wondertree/posecamera --images ./tmp/female_pose.jpg --cpu
qt.qpa.xcb: could not connect to display :1
qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in "/home/user/miniconda/lib/python3.8/site-packages/cv2/qt/plugins" even though it was found.
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.

Available platform plugins are: xcb.

Hi again,

I am working with your library nowadays. When I work with an image containing multiple people, it only gives value to a single person key points. For example:

 det = posecamera.pose_tracker.PoseTracker()

image = cv2.imread("/content/1.png")

pose = det(image)
for name, (y, x, score) in pose.keypoints.items():
    #cv2.circle(image, center_coordinates, radius, color, thickness)
    cv2.circle(image, (int(x), int(y)), 8, (243, 255, 51), -1)

img = cv2.resize(image, (250, 250))        
cv2_imshow(img)

How can i get it for multiple person? Thank you.

Description

PyPi package integrated and documentation updated to the latest version.

Resolved Issues

Closes #6 PyPi python package Closes #7 Github CI on pull request

Thank you very much for creating this module.

is there a way to get the keypoints as pixels numbers ? Because right now it looks like this :

{'nose': array([-5.9867120e+00, 9.0966003e+02, 3.6360759e-01], dtype=float32), 'left_eye': array([-4.8970135e+01, 9.4581982e+02, 9.4886214e-02], dtype=float32), 'right_eye': array([-5.1433197e+01, 8.6966357e+02, 9.2111975e-02], dtype=float32), 'left_ear': array([8.2545176e+00, 9.7673682e+02, 4.2675975e-01], dtype=float32), 'right_ear': array([2.1643393e+01, 7.6527582e+02, 5.2597636e-01], dtype=float32), 'left_shoulder': array([3.0106079e+02, 1.0566696e+03, 6.5383315e-01], dtype=float32), 'right_shoulder': array([3.2392834e+02, 6.6651459e+02, 4.0876174e-01], dtype=float32), 'left_elbow': array([5.6922583e+02, 1.3786616e+03, 6.2058413e-01], dtype=float32), 'right_elbow': array([6.7028491e+02, 6.0433405e+02, 5.2336699e-01], dtype=float32), 'left_wrist': array([8.1140930e+02, 1.1170122e+03, 5.4489833e-01], dtype=float32), 'right_wrist': array([1.01784973e+03, 5.69821289e+02, 7.70775676e-01], dtype=float32), 'left_hip': array([1.0433268e+03, 9.6914252e+02, 6.5443105e-01], dtype=float32), 'right_hip': array([1.0465645e+03, 7.2657220e+02, 7.1066779e-01], dtype=float32), 'left_knee': array([1.6087389e+03, 1.0228279e+03, 7.9422629e-01], dtype=float32), 'right_knee': array([1.5593778e+03, 7.6752869e+02, 7.4263895e-01], dtype=float32), 'left_ankle': array([2.0873940e+03, 1.2951686e+03, 7.3480493e-01], dtype=float32), 'right_ankle': array([2.1593096e+03, 8.6261426e+02, 7.3734212e-01], dtype=float32)}