This example requires to install two additional libraries for rendering

pip install pyrender --user
pip install trimesh --user

If you are running the script on a remote machine via SSH, then might run into the following error:

ValueError: Failed to initialize Pyglet window with an OpenGL >= 3+ context. If you're logged in via SSH, ensure that you're running your script with vglrun (i.e. VirtualGL). The internal error message was ""

To fix this, you need to run the PyOpenGL in a headless configuration which is not enabled by default. Just uncomment the following line(4) in discover_latent_keypoints.py

# os.environ["PYOPENGL_PLATFORM"] = 'egl'

This will use the GPU accelerated rending on your remote machine. To use CPU-accelerated rendering, your need to use OSMesa instead of EGL. However, this is not tested yet.

# os.environ["PYOPENGL_PLATFORM"] = 'osmesa'

If you run into the following error:

tensorflow.python.framework.errors_impl.UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above. [Op:Conv2D]

Either you are short on memory (reduce batch size) or uncomment the following line(5) in discover_latent_keypoints.py

# os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true'