The Blur.me team is developing real-time live image processing technology and existing image blur processing.
Processing a video instead of an image requires a lot of computation. Therefore, advanced technologies such as deep learning and image compression are needed. In addition, to process it seamlessly in real-time, to achieve high performance, it must be implemented by utilizing hardware such as GPU. The following technical elements must be implemented for it to be effective.
The Blur.me team is working hard to develop algorithms that can mask images with a performance of 30 frames per second or more. Our team is implementing this to work on essentially any hardware, even on Nvidia's Jetson Nano boards. It's known to be low-spec hardware priced at 100,000 won, which isn't considered a high-performance deep learning machine.
It is still under development, but it will be widely used in CCTV and real-time broadcasting when it is completed.
Below is a demo video.