Deep Learning-based Super Resolution IP

c.WAVE120 is a deep learning-based super-resolution HW IP that upscales low-resolution data into high resolution in real-time. The neural network on c.WAVE120 is designed and trained to upscale video horizontally and vertically to two, three, and four times larger with enhanced resolution results. c.WAVE120 performs the upscaling tasks through utilizing a massive set of training datasets. c.WAVE120 extracts the feature points of an image or video, splits them pixel by pixel, applies the appropriate colors to fill in the missing parts of the data, stitches them, and then reproduces sharper, higher-resolution images.

In modern technology, neural networks are implemented in the form of processors or accelerators of processors. However, this type of implementation tends to be hard to optimize performance due to structural limitations.