NAOqi PeoplePerception - Overview | API | Tutorials
ALFaceDetection is a vision module in which the robot tries to detect, and optionally recognize, faces in front of him.
ALFaceDetection is based on a face detection/recognition solution provided by OMRON.
Face detection detects faces and provides their position, as well as a list of angular coordinates for important faces features (eyes, nose, mouth).
To make the robot not only detect but also recognize people, a learning stage is necessary. For further details, see: Learning stage for recognition.
Recognition feature returns for every image the names of people that are recognized.
Temporal filter: in addition, there is temporal filter output to easily build higher level features using recognition. Indeed we don’t want the robot to say “Hello Michel” several times per second, so someone’s name will only be output the first time he is recognized and will be placed in a short term memory. This memory will be kept as long as some faces is not only recognized but detected by the robot. As soon as there are more than 4 seconds without detecting any face, the short term memory is cleared and Michel name will be output again if the robot encounters him. This is that output that is used in the Choregraphe Face Reco box.
Once ALFaceDetection is started, the event FaceDetected() returns a value organized as follows:
FaceDetected =
[
TimeStamp,
[ FaceInfo[N], Time_Filtered_Reco_Info ],
CameraPose_InTorsoFrame,
CameraPose_InRobotFrame,
Camera_Id
]
TimeStamp: this field is the time stamp of the image that was used to perform the detection.
TimeStamp =
[
TimeStamp_Seconds,
Timestamp_Microseconds
]
FaceInfo: for each detected face, we have one FaceInfo field.
FaceInfo =
[
ShapeInfo,
ExtraInfo[N]
]
ShapeInfo: shape information about a face.
ShapeInfo =
[
0,
alpha,
beta,
sizeX,
sizeY
]
ExtraInfo: shape information about a face.
ExtraInfo =
[
faceID,
scoreReco,
faceLabel,
leftEyePoints,
rightEyePoints,
unused, # for backward-compatibility issues
unused,
nosePoints,
mouthPoints
]
EyePoints =
[
eyeCenter_x,
eyeCenter_y,
noseSideLimit_x,
noseSideLimit_y,
earSideLimit_x,
earSideLimit_y,
always_zero, # for backward-compatibility issues
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero
]
NosePoints =
[
bottomCenterLimit_x,
bottomCenterLimit_y,
bottomLeftLimit_x,
bottomLeftLimit_y,
bottomRightLimit_x,
bottomRightLimit_y
]
MouthPoints =
[
leftLimit_x,
leftLimit_y,
rightLimit_x,
rightLimit_y,
topLimit_x,
topLimit_y,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero,
always_zero
]
Time_Filtered_Reco_Info can be equal to:
CameraPose_InTorsoFrame: describes the Position6D of the camera at the time the image was taken, in FRAME_TORSO.
CameraPose_InRobotFrame: describes the Position6D of the camera at the time the image was taken, in FRAME_ROBOT.
Camera_Id: gives the Id of the camera used for the detection (0 for the top camera, 1 for the bottom camera).
Performances
Limitations
Performances
When learning someones face, the subject is supposed to face the camera and to keep a neutral face because a neutral face is between sadness and happiness. Otherwise, it would be harder to recognize someone sad if he was smiling during the learning process.
Sometimes, depending on a change of location or haircut, a known face can be difficult to recognize. To improve the robustness, a reinforcement process as been added. If someone is not recognized, or mistaken for someone else, just learn him again. This learning will be added to that person’s database. After some days, you should get more reliable recognitions.
Limitations
Recognition is less robust than detection regarding pan, tilt, rotation and maximal distance. Reason is that the recognition algorithm doesn’t have a 3D representation of the person to recognize and uses some info like distances between keypoints for the recognition. If we turn the head, distances ratios will be modified.
Performances
The learning stage takes five consecutive images and will try to learn a user’s face from each of these images.
Limitations
The learning stage will only consider the bigger face found in the field of view.
To get a feel of what the ALFaceDetection can do, you can use Monitor and launch the vision plugin. Activate the face detection checkbox and start the camera acquisition. Then, if you present your face to the camera - or show a picture with a face on it - Monitor should report the detected faces with blue crosses.
Another way to use face detection is to launch the Choregraphe Face Tracker box. The robot will try to keep a detected face in the middle of its field of view.
Learning stage can be done via the learnFace bound method of the API or through user friendly interface of Choregraphe Learn Face box.
Note
The algorithm requires better conditions for the learning stage than the ones needed for detection.
Note
You can launch the Face Tracker box in parallel with the learning stage so the face to learn will always be in the middle of the robot’s field of view.