Although it is possible to run DeepFace in parallel, it depends on TensorFlow, and since it allocates all CPU cores to a single task, running multiple instances in parallel becomes ineffective as the cores are shared between different workers.