Speculo: Pattern Recognition with a Deep Convolutional Inverse Graphics Network for Face-Indexing

Footage analysis procedures require manual examination which can be time consuming. As an alternative to the above procedure, we propose a system capable of automating footage analysis by fingerprinting all identified faces and classifying them into unique faces, regardless of the angle the face was found in the footage. This system also has the ability to expand the number of faces it can recognize during runtime opposed to detecting a fixed number of faces.