Face Recognition

Scaled Recognition provides state of the art face recognition as part of our API.  In Figure 1 below you can see how we've extracted the faces and then matched them against a user's social graph, subsequently extracting relevant metadata about the user.  

 

Fig. 1.  Operational face recognition and Facebook graph matching.


Figure 2 shows the multi-step processing pipeline that all input images experience.  This pipeline excludes the initial "Training" of the model, which requires that we ingest a set of images and train a neural network on them.  The combination of these steps is complicated and computationally intensive.  You don't want to try this at home if you can help it!



Fig 2. Operational recognition pipeline

In some cases, users are interested in parts of our pipeline, such as image extraction.  We can provide access to these intermediate stages through our API.

Comments