+1! There's such a wealth of computer vision approaches of practical value locked up in CVPR and other research papers that could be exposed this way to enthusiasts and semi-pro users