Lineups
Face detection
- Quickly detects multiple faces in images.
- Includes an advanced tracking feature that enables even faster detection for continuous images (cut-out video frames).
Face parts detection
- Detects the coordinates of facial features such as eyes, nose and mouth
Age estimation
- Estimates the generation of the person in the image.
Eye direction recognition
- Detects eye direction (eye balls orientation).
Blink recognition
- Estimates the degree of eye opening and closing.
Face recognition
- Recognizes identity by matching the detected face against a pre-registered face (more than one face view per person should be registered for recognition).
- Performs 1-to-1 matching and 1-to-N matching.
Smile recognition
- Estimates the degree of smiling.
Gender recognition
- Estimate the gender of the person in the image.
Face direction recognition
- Estimates the face direction (angle).
Features
Rich functionality
- Face detection, face recognition, age estimation, gender estimation, etc.
Portability
- Can be ported to various types of platforms. Customizable according to memory size and CPU power. (Option)
Applicable to visible light camera and near infrared camera
- Near infrared camera can be used as well as normal camera (visible light camera).
Optimization of accuracy
- Optimization of accuracy is available according to user application and use case. (Option)
Use case
Driver Monitoring System
- Detect driver's face and monitor eye direction, face direction and head position.
Camera Control
- Automatically focus on the registered face.
Robot Control
- Detect human face and take action.
Specification
| FaceU Version 2.6 Enbedded version（reference value） | ||
|
Size of software Face detection/ Face parts detection/ Face recognition |
ROM | 940 KBytes |
| RAM | 1 KBytes | |
| WORK |
58 MBytes |
|
| Size of dictionary |
16 MBytes |
|
| Input image | Size | MAX 8192 x 4320 or 4320 x 8192 |
| Format | 8 bit gray scale | |
| Number of face parts detection | Max 83 points | |
| Condition of face recognition |
Distance between eyes must be more than 20 pixels（≒face width must be more than 40 pixels） Can not recognize the face with mask or sunglasses |
|
|
Processing speed （Renesas RCarE2 ARM CoretexA7-1.0GHz） |
Face detection | 40 msec |
| Face parts detection |
138 msec（83 points） |
|
|
Characteristic point extraction for face recognition |
146 msec（0.01 msec for 1 to 1 matching） | |
| Others | Contact us for PC and Android version | |
Operability confirmed SoC / EVA board
R-CarM2 Application Development Board
Renesas Starter Kit ＋ for RZ/A1H
AtmarkTechno
Armadillo-810
mbed™Board
GR-PEACH Full
Raspberry Pi2,
Panda Board, etc.