concept face recognition in category deep learning

This is an excerpt from Manning's book Deep Learning for Vision Systems MEAP V08 livebook. Login to get full access to this book.

Let’s look at some practical day-to-day information retrieval algorithms that utilize the concept of visual embeddings. Some of the prominent applications to retrieving similar images given an input query, include: face recognition (FR), image recommendation, and object re-identification systems.

Figure 10.3: Illustrative face verification and face recognition systems. On the left, an example of a face verification system comparing one-on-one matches to identify whether the image is for Sundar or not. On the right, an example of face identification system comparing one-to-many matches to identify all images. Despite the objective-level difference between recognition and identification, they both rely on a good embedding function that captures meaningful differences between faces. Figure is inspired from the blog post by Satya Mallick on face recognition.