There are multiple distinctions when considering why a face is not detected. The face must first be used, recognized, detected, then identified.
Faces that do not meet a minimum size and quality criteria become unused if there is no good quality image of these faces in the library to create a model. When this happens, these faces are not displayed within the "Detailed Information" page.
The minimum size and quality of an image must be at least 200px by 200px to be recognized. Note that the face itself should be 200px by 200px, not just the image itself.
Unrecognized faces occur when the AI can identify a face in the image, but due to there not having a model set in place to compare the face with, the face is unable to be recognized. This means that although the face may appear in the "Detailed Information" section, they will not have a card in the People Dashboard as a model as not yet been created of them yet.
For them to be recognized, a model must be created with a high quality image at least 200px by 200px in size as mentioned before.
Unrecognized faces will appear in the "Detailed Information" section in Face Recognition 1.2 and newer versions.
If a face is obstructed or covered, it may lead to complications with Face Recognition detecting if there is a face in the image. An undetected image is simply an image where a face in the image is not picked up by the AI.
To be detected, the face should be unobstructed as much as possible to ensure that facial features and other key factors in Face Recognition can be utilized.
An unidentified face means that the image is detected and recognized, but it has not yet been tagged with a name. Unidentified faces can be found in the "Detailed Information" section of an asset or in the "Unidentified" tab in the People Dashboard. From there, you can easily tag the unidentified images by following these steps.