I’ve written a Jupyter notebook application to capture video from a camera and perform face detection and display it on the jupyter notebook using gstreamer framework with Python bindings. However, the video is very slow because I’ve to convert that into jpeg before displaying it on the notebook.
How can the performance be improved so that it will be close to real time?
How can I display it without converting into jpeg. I’m using BGR format.
In C++ based gstreamer applications, it opens OpenGL renderer which supports wide range of formats. How do we integrate that in jupyter notebook if that method can improve performance?
Please guide regarding this.