Pose estimation using MoveNet on videos

[hamza naeem]

I am trying to pass the video to movenet model for pose estimation, I have successfully run the model on image but I am stucked on passing video to it . How can we pass the video like mp4 extension to it .

Notebook link:

 https://github.com/tensorflow/hub/blob/master/examples/colab/movenet.ipynb

[Morgan Roff]

You will need to extract the individual frames from the video, and pass those to the MoveNet model. Within TensorFlow, the tfds.features.Video feature connector might help. If that doesn't meet your needs, I expect there are other Python libraries that can help you retrieve frames from videos as images.

I hope that helps.

[hamza naeem]

Can you please view the notebook and change the respective changes you think please