Page tree

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

Version 1 Next »

Short explanation for terms that are found within visage|SDK documentation and in face tracking in general.

General terminology

  • Action units – fundamental actions of individual muscles or groups of muscles, estimated and returned by the visage|SDK (e.g. lower lip drop, right outer brow raiser). Often applied to a 3D object’s Morph targets

  • Age estimation / Age classification – estimation of a person’s age in a frame or a continuous stream of frames

  • Emotion estimation / Emotion detection– estimation of intensities of human emotions from a predefined set in a frame or a continuous stream of frames

  • Eye tracking – localization of pupil positions in a continuous stream of frames

  • Face detection – detection of face bounding boxes in a frame

  • Face filter / AR face filter / Facial mask – a mask-like augmented reality that adds virtual objects to an individual's face

  • Face identification / Registered Face Recognition – process of determining person’s identify by checking its face descriptor against a database of labeled face descriptors

  • Facial occlusion – an obstructed view of the face where only parts of the face are visible. Obstruction can be person’s hands, glasses, mask, beard, etc…

  • Face recognition – extracting and matching face descriptors from a frame

  • Face template / Face descriptor – unique identifier of the human face, usually represented as an array of values

  • Face tracking – tracking/localization of feature points in a continuous stream of frames

  • Face verification / Registered Face Recognition – process of verifying if two face descriptors match i.e. belong to the same person

  • Facial landmarks / Feature points – salient points on the human face

  • Gender estimation / Gender classification – binary estimation of person’s gender in a frame or a continuous stream of frames

  • Head tracking – estimation of head pose in a continuous stream of frames

  • Morph target / Blend shapes / Shape keys – a variation of the base mesh. Morph targets are typically used for facial animation. The base mesh defines the neutral expression and the morph targets define expression such as "smile", "frown", "eyes closed"

  • Morph target animation - animating a 3D object using morph targets. When applied to a human face, for example, the head is first modeled with a neutral expression and a "target deformation" is then created for each other expression.

  • Multiple-face detection – detection of more than one face bounding boxes in a single frame

  • Multiple-face tracking – tracking/localization of feature points of multiple faces in a continuous stream of frames

Automotive related terminology

  • Driver monitoring system / DMS – a camera-based system pointed at the driver’s face which provides a real-time evaluation of the presence and the state of the driver

  • Occupant monitoring – Monitoring of all passengers in the vehicle to better understand their state and condition

  • Driver drowsiness monitoring – monitoring the level of fatigue in the driver (eg by monitoring eye closure, blinking, yawning…)

  • Driver identification – Verifying the driver's identity, eg by using facial recognition, usually in order to provide access to specific car functions

  • Driver assistance – the function of the car that allows it to take control of a least one significant car function from the driver when necessary

  • No labels