Overview
The face image assessment module provides metrics including head pose, smile detection, inter-eye distance, closed eyes, etc.
Face image quality issues to investigate:
Input
-
BQAT-CLI: Create a folder named
data
under your working directory and put your face images in this folder. -
BQAT-API / BQAT-Stateless: Upload images via web API.
-
BQAT-GUI: Upload images via web page interface.
Supported file type: JPG, JPEG, JP2, BMP, PNG
Output
BQAT will produce quality metrics generated by processing engine selected in addition to input file metadata. The outputs will be stored as CSV
via BQAT-CLI or JSON
via BQAT-API.
Optional processing engines:
- BQAT (native)
- OFIQ (beta)
- BIQT (beta)
BQAT:
Column | Description |
---|---|
file | Path to the input file |
ipd | Inter-pupillary distance |
confidence | Confidence level of face dectection (not quality score) |
bbox_left | Left border of the face bounding box coordinates in pixels |
bbox_upper | Upper border of the face bounding box coordinates in pixels |
bbox_right | Right border of the face bounding box coordinates in pixels |
bbox_bottom | Bottom border of the face bounding box coordinates in pixels |
eye_closed_left | Left eye close or not |
eye_closed_right | Right eye close or not |
pupil_right_x | X coordinates of right pupil in pixels |
pupil_right_y | Y coordinates of right pupil in pixels |
pupil_left_x | X coordinates of left pupil in pixels |
pupil_left_y | Y coordinates of left pupil in pixels |
yaw_pose | Yaw in head pose direction |
yaw_degree | Yaw in head pose degree |
pitch_pose | Pitch in head pose direction |
pitch_degree | Pitch in head pose degree |
roll_pose | Roll in head pose direction |
roll_degree | Roll in head pose degree |
smile | Smile detected or not |
glasses | Glasses detected or not |
image_width | Width of the input image in pixels |
image_height | Height of the input image in pixels |
OFIQ:
OFIQ engine is still in early stage of development, might be unstable, use with care.
Column | Description |
---|---|
file | Path to the input file |
quality | Overall image quality score |
background_uniformity | Image background uniformity |
illumination_uniformity | Image illumination uniformity |
luminance_mean | Mean of image luminance |
luminance_variance | Variance of image luminance |
under_exposure_prevention | Under exposure prevention |
over_exposure_prevention | Over exposure prevention |
dynamic_range | Image dynamic range |
sharpness | Image sharpness |
compression_artifacts | Artifacts detected for image compression |
natural_colour | Naturalness of image colour |
single_face_present | Probability of single face presented |
eyes_open | Probability of open eyes detected |
mouth_closed | Probability of closed mouth detected |
eyes_visible | Probability of visible eyes |
mouth_occlusion_prevention | Mouth occlusion prevention |
face_occlusion_prevention | Face occlusion prevention |
inter_eye_distance | Inter eye distance |
head_size | Head size |
leftward_crop_of_the_face_image | Left of the face crop |
rightward_crop_of_the_face_image | Right of the face crop |
downward_crop_of_the_face_image | Bottom of the face crop |
upward_crop_of_the_face_image | Top of the face crop |
head_pose_yaw | Yaw of head pose in degree |
head_pose_pitch | Pitch of head pose in degree |
head_pose_roll | Roll of head pose in degree |
expression_neutrality | Face expression neutrality |
no_head_coverings | Probability of no head coverings |
BIQT:
Column | Description |
---|---|
file | Path to the input file |
background_deviation | Image background deviation |
background_grayness | Image background grayness |
blur | Overall image blurriness |
blur_face | Face area blurriness |
focus | Overall image focus |
focus_face | Face area focus |
openbr_IPD | Inter eye distance from openbr |
openbr_confidence | confidence value from openbr |
opencv_IPD | Inter eye distance from opencv |
opencv_eye_count | Eye count from opencv |
opencv_face_found | Face count |
opencv_face_height | Height of face detected |
opencv_face_width | Width of face detected |
opencv_frontal_face_found | Number of front facing head found |
opencv_landmarks_count | Landmarks of face detected |
opencv_mouth_count | Number of mouth detected |
opencv_nose_count | Number of nose detected |
opencv_profile_face_found | Number of side profile of head |
over_exposure | Overall image exposure value |
over_exposure_face | Face area exposure value |
quality | Overall quality score |
sap_code | Sap code |
skin_ratio_face | Skin to face area ratio |
skin_ratio_full | Skin to fill image area ratio |
image_area | Image area |
image_channels | Number of image colour channels |
image_width | Width of the input image in pixels |
image_height | Height of the input image in pixels |
image_ratio | Image aspect ratio |
openbr_left_eye_x | Left eye x coordinate in pixels |
openbr_left_eye_y | Left eye y coordinate in pixels |
openbr_left_eye_x | Right eye x coordinate in pixels |
openbr_left_eye_x | Right eye y coordinate in pixels |
opencv_face_center_of_mass_x | Face center of mass x coordinate in pixels |
opencv_face_center_of_mass_y | Face center of mass y coordinate in pixels |
opencv_face_offset_x | Face offset x coordinate in pixels |
opencv_face_offset_y | Face offset y coordinate in pixels |
opencv_face_x | Face x coordinate in pixels |
opencv_face_y | Face y coordinate in pixels |
opencv_left_eye_x | Left eye x coordinate in pixels |
opencv_left_eye_y | Left eye y coordinate in pixels |
opencv_right_eye_x | Right eye x coordinate in pixels |
opencv_right_eye_y | Right eye y coordinate in pixels |
opencv_mouth_x | Mouth x coordinate in pixels |
opencv_mouth_y | Mouth y coordinate in pixels |
opencv_nose_x | Nose x coordinate in pixels |
opencv_nose_y | Nose y coordinate in pixels |
Not all the columns are included in the table above for simplicity, for instance, there are normalized or scalar value of the same metrics.