This is a live demo of our current work in automatic visual machine learning. Visual machine learning seeks to bridge the semantic gap by analyzing the pictorial content, the pixels, and giving a human understandable description. These technologies are widely applicable in areas ranging from searching image databases and computer vision, to computer aided diagnosis, web searching, forensic science and satellite image analysis.
We are currently investigating new architectures in deep learning with Convolutional Neural Networks (CNNs) and so the underlying CNN architecture will regularly change. For specific questions, please feel free to contact us - see below.