We’re a team of former Imperial College post-doc researchers who are looking to make some of our state-of-the-art algorithms available via an API. We’re not simply sticking OpenCV on the end of a REST API here, but attempting to make the capabilities of a team with 20+ years of computer vision expertise accessible to anyone.

Some of the proprietary functionality we’re currently looking to provide includes:

Super Resolution (Video Resolution Upscaling) - Our algorithm extends recent medical imaging research and is able to convert low resolution videos to high resolution ones whilst minimising loss of quality. Our methods can achieve much better results than interpolation based approaches (example: https://www.dropbox.com/s/74duixpek7fxhl4/superResolution.png?dl=0).

Image/Video Depth of Field Adjustment - Our algorithm is able to estimate depth information from and adaptively apply filters to emulate different depth of field - effectively allowing for a clip captured with a smartphone to look like it was shot with a different lens (example: https://www.dropbox.com/s/7qhfgnwl08vtk63/compare.png?dl=0).

We’re intending to eventually make as many state-of-the-art computer vision algorithms available as possible and would love your feedback on the kinds of image/video processing tasks that you would want to be able to use.

Would you find this a useful tool? If so, what functionality would be most valuable for you?