Nandita Mathur, chief strategy officer and head of engineering, at Q3 Technologies, a biometric facial recognition solutions provider, says the technology has undergone vast improvements. We first capture full-attention images of persons whom we are going to recognize through facial recognition system,” she explains. “A database is then created with face encodings of the images [128 landmark points are used for face encodings].” With real-time video stream, frames are captured and are ingested in a machine learning algorithm, Mathur says. “In every frame, the algorithm recognizes faces and extracts individual face images for further analysis,” she says.