By detecting whether there is someone in the current camera preview and detecting the orientation of the human face (head), important information can be provided to the decision-making system of the smartphone, for example, in scenarios such as intelligent screen-on, intelligent screen-off, intelligent rotation, and image rotation control. Face orientation refers to the direction to which the central base line (pointing to the top of the head) of the human head faces on the facial plane, and includes five scenarios: upward, rightward, downward, leftward, and no direction (no person).
Face orientation recognition identifies face (head) orientation in an image, and returns the orientation information, including upward-facing, right-facing, downward-facing, left-facing, and no face, by utilizing advanced image recognition technologies. The API obtains the specific face orientation category and degree of confidence.
This algorithm detects face (head) orientation information in images, and can be applied to scenarios in which the app needs to determine whether there is a person in the image, as well as the person's face orientation. Example:
The closer the confidence score is to 1, the more reliable the recognition result. If the recognition result of a 640 x 480 image shows that the headpose is upwards, and the confidence score is 0.99972343, it means that the result is very reliable.
For details, see Supported Devices.
For details, see Supported Countries/Regions.
Function |
Restrictions |
---|---|
Input image size |
Requirement on the image size: height/width = 3/4. Otherwise, error code 200 is returned, indicating that the input parameter is invalid. |
Multi-thread calling |
Multi-thread calling is supported. |
Intra-process and inter-process support |
Currently, the face orientation detection engine supports both intra-process and inter-process modes to suit the needs in different scenarios. |
App version |
This API can be called only by 64-bit apps. |