• Using a 3-D model of a 'average' forward-looking face, the DeepFace algorithm first aligns a face so that the person in the photo faces forward. The deep learning is then used to generate a numerical description of the forward-looking face.
• DeepFace determines that two photos with comparable enough descriptions must depict the same face.
• DeepFace, according to Facebook, is currently only a research project. The algorithm could theoretically be expanded to cover a considerably bigger number of identities before being put to Facebook's social network.

Read More: What is Skype translator?