Enhancing face:background_job Performance through GPU Parallelization

[stanke-sk]

Hello Everyone,

First off, I want to express my gratitude to everyone who has contributed to this project. I have a question regarding the possibility of parallelizing the face:background_job on a GPU. I've noticed that my GPU utilization is quite low,nearly idling, and it's only using about 1344MB of memory with the analysis_image_area --value 6291456 setting as the highest allowed by the model 4. I'm looking to enhance the image processing speed further. I have the capability to employ even additional GPU cards if necessary. Is there a way to achieve this, or is it something that could be considered for future development? I'm eager to maximize GPU or GPUs' capabilities. Thank you!

[matiasdelellis]

There is a lot of progress in parallelism, but regarding the use of GPU you must compile dlib and pdlib yourself to take advantage of it.

ecd603a