Patent attributes
In general, this disclosure describes techniques for assessing image quality of captured facial images. An example method includes capturing an image, generating a facial detection confidence score based in part on a likelihood that a representation of at least a portion of a face is included in the image, generating a facial landmark detection confidence score based at least in part on a likelihood that representations of facial landmarks are accurately identified in the image, and generating a geometric consistency score based at least in part on a difference between a point of intersection between a nose base and a line segment that passes through each eye and a midpoint of the line segment. The method also includes generating an image quality score based in part on a combination of the confidence scores, and the consistency score, and classifying an image quality based on the image quality score.