Now use pyqt develop an interface under Windows, one of the function to play video, and invoke existing machine learning algorithm of video processing,
Currently used QLabel, each frame for processing and display, call machine learning algorithms of processing data part has set up a child thread to do it, but still be caton,
Algorithm part and call tensorflow algorithm is used, and the thread is not set delay (sleep),
Would like to ask this kind of circumstance is the problem of the GPU, or is the issue of multithreading (how much the entire software is a thread)?

Ask more, here, python multithreading and processes will be a difference? Will not use multiple processes better?
Thank you all!