Page 1
White Paper
Issued By:Sensing & Module Division
Issue Date: May 2018
Image Sensor for Human Recognition
Human Vision Components, “HVC-P2”
Introduction
OMRON Corporation has been developing image sensing technology “OKAOTM Vision”, for Human Recognition for over 20
years, which enables the detection of faces, facial features, human bodies coupled with the estimation of age or gender of
the humans in the captured images. Omron had licensed this technology into mainly digital cameras and smartphones, and
consequently those have amounted to a total over 10 million licenses (*1). Furthermore, OMRON has launched Human
Vision Components (commonly referred to as HVC) , which are modular solutions, and each incorporates 10 human sensing
functions which are features of “OKAOTM Vision”, including the functions mentioned above.
*1: The number of licenses for total agreements.
Human Vision Components “HVC-P2”, Image Sensor for Human Recognition
Human Vision Components (HVC), the image sensor for Human Recognition, is a combination of Omron’s own image
sensing technology “OKAOTM Vision”, which can recognize people’s blinks, gaze direction, expression, gender or age, with a
camera module.
Response time has become 10 times faster than the first generation , HVC-P, at maximum rate.
Moreover, there are two types of lens options available, a long-distance 50 degree type (B5T-007001-010) and a wide-angle,
90 degree type (B5T-007001-020). This allows customers to make a choice depending on detection distance and FOV(Field
of Vision). “HVC-P2” is able to detect and estimate the attributes or condition of the objects without notifying them of the
presence of its camera, and realizes the optimal service which corresponds to users’ needs.
Long-distance type (B5T-007001-010)
• This product can detect a human body as far as 17m, and also a face at a
maximum distance of 6m.
• This product is for detection of human bodies or faces which exist in a little
far away from the device.
Wide-angle type (B5T-007001-020)
• This product can detect human bodies and
faces when they exist in the range between +90°and -90°horizontally.
• This product is for human body or face
detection for a short distance and wide FOV.
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
1
Page 2
White Paper
The Installed Functions
• Human Body Detection,
• Face Detection,
• Hand Detection
It can output the sizes and position coordinates of the human bodies,
faces or hands which are captured in the images.
• Age Estimation
• Gender Estimation
• Expression Estimation
• Face Direction Estimation
• Gaze Estimation
• Blink Estimation
It can analyze the facial data gathered from the images
by its own algorithm and output the results respectively.
• Face Recognition
It conducts a matching check between the people who are captured
in the images and the facial data (the amount of features) which had
been inputted in advance to the main memory and outputs the results.
Development background of this product
The need for human recognition sensors has been accelerated recently to join in the development of analysis technology, coupling with
advanced computing technology in the IoT market such as deep learning (machine learning). In addition, there has been an expansion in
use and an increase in the objects for sensing; hence the needs have become more and more diverse. Moreover, under these circumstances,
the needs for grasping the locations, attributes, or condition of end users become increasingly important in advanced analytics.
OMRON has been offering its own image sensing technology, “OKAOTM Vision”, through its software library to the manufacturer of
assembled products, and its technology has been installed mainly to the digital devices, such as digital cameras and smartphones.
On the other hand, we have realized that this software has place in other applications in which a user does not have the processing to run
an algorithm. Therefore, we have developed the new HVC-P2 which have become closer to hardware so that the customers can integrate
OMRON image sensing technology into their system and bring their product to market faster.
The main board is compact and it can detect the human features up to four times per second by taking advantage of the features of
“OKAOTM Vision” which enables high speed processing.
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
2
Page 3
White Paper
Installation examples
Digital Signage
• For an interactive experience which corresponds to the customers
who stand in front of the digital signage. (Face Detection, Age
Estimation, Gender Estimation)
• Utilization in the audience measurement and the advertising
effect measurement by gathering the customers’ attributes data,
such as age and gender, who see the digital signage. (Face
Detection, Age Estimation, Gender Estimation)
• Displaying an operation panel for content or touchscreens at the
height of the person’s face, which is estimated by the facial
position data assembled by the machine. (Face Detection)
Vending machine
• Utilization in marketing activities by gathering the customers’
attributes data, such as age and gender, and connecting them
with the products purchased.
• Displaying the recommendable products corresponding to the
customers’ age or gender who stand in front of the machine.
(Face Recognition, Age Estimation, Gender Estimation)
About the offering of the developing tool
Omron is offering the following developing tools for free to allow for ease of evaluation, in hope that this may accelerate
development and materialize proof of concepts within a short term. Please visit our website for details.
https://www.components.omron.com/mobile/hvc_p2?nodeId=40701010
Evaluation Software
• Corresponding to Windows 7 / 8.1 / 10, 32/64bit version
• Software which can be used handily on your personal computer and let you experience the basic functions and
performance of the HVC-P2.
USB driver
• For Windows 7 / 8 / 8.1 (32/64 bit)
• If you operate the programs on Windows OS, you do not need to specially prepare another driver.
*If you are using Linux OS, you can use the driver attached as standard equipment.
Sample Code
• Currently C and Python languages are available.
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
3
Page 4
White Paper
Function
Output
Note
1000, by 1
higher
number indicates higher
About the offering of the developing tool Specifications
Item
Specifications
B5T-007001--010 B5T-007001-020
Detection resolution
Horizontal detection range
(angle of view)
Vertical detection range
(angle of view)
1600×1200 pixel 1600×1200 pixel
54° 94°
41° 76°
Item Specifications
Output Image
Image format RAW (8bits, Y data)
• Number of the objects detected
Human Body Detection
Face Detection
Hand Detection
• Detected location
(position coordinates)
• Detected size
• Reliability
<Options> No output / 160×120 pixel / 320×240 pixel
• Maximum 35 objects
• Coordinates (in pixels) from the top-
left corner of the screen
• The pixel size on the screen
• Reliability of the results (0 -
degree, larger number indicates
reliability)
• Horizontal angle
Face Direction Estimation
Gaze Estimation
Blink Estimation • Blink degree
• Vertical angle
• Tilt angle (roll)
• Reliability
• Horizontal angle
• Vertical angle
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
• Angle (in degrees)
• Reliability of the results (0 - 1000, by 1
degree, larger
reliability)
• Angle (in degrees)
• Output for both of the eyes
• (1 - 1000, by 1 degree,
• When the eye is closed tightly, the
degree of confidence will become
larger.
4
Page 5
White Paper
Function
Output
Note
the result of
About the offering of the developing tool Specifications
Age Estimation
Gender Estimation
Expression Estimation
• Age
• Reliability
• Gender
• Reliability
• Scores for 5 expressions
• Positive / Negative degree
• 0-74, for those over 75, estimated by 1 year.
• Reliability of the results
(0-1000, by 1 degree,larger
number indicates higher reliability)
• Male / Female
• Reliability of the results
(0 - 1000, by 1 degree, larger number indicates higher
reliability)
• The degree of each expression (“neutral”, “happiness”,
“surprise”, “anger”, “sadness”,
0 – 100, by 1 degree,the larger number indicates the higher
likeliness of the expression)
• +100 -－100, by 1 degree, stronger “happiness” would
achieve near +100 score, and
if the other expressions, such as “surprise”, “anger”,“sadness”
are stronger, the “happiness” score would be near －100.
Face Recognition
• Results of individual
identification
• Score
• Outputs either information,
User ID of the person specified from the registered data or
non-registered message.
• Maximum registered profiles: 100
• 10 data per person can be registered
• Facial images photographed by other cameras cannot be
registered
• Matching degree (0 – 1000, by 1 degree),outputs
a user who achieved the highest matching degree,the
number close to 1000 indicates the high probability of
correspondence
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
5
Page 6
White Paper
results
About the offering of the developing tool Specifications
■Rating
Item Rated Value Item
Power Supply Voltage
Current Consumption Under 0.4A Current Consumption
Operating Temperature
Operating Humidity
DC5V±10%
0～+50℃
(without condensation or freezing)
Under 90%RH
(without condensation or freezing)
Power Supply Voltage
Operating Temperature
Operating Humidity
■Detection Distance (*reference value)
Function
B5T-007001-010 B5T-007001-020
Human Body Detection 17m 8m
Hand Detection 6m 3m
Face Detection 10m 5m
Face Direction Estimation,
Gaze Estimation,
Blink Estimation,
Age Estimation,
Gender Estimation,
Expression Estimation,
Face Recognition
3m 1.5m
Maximum Distance
*When attempted object detection distance is further than the above values, the level of accuracy will be lowered. In addition,
the detectable distance will vary, dependent on component settings.
■Communication with Host (for both B5T-007001-010 and B5T-007001-020)
＜UART＞
Outline
Communication Protocol Non-procedural
Synchronization Scheme Asynchronous method
Data format Start: 1 bit, Data: 8 bit, Stop: 1 bit, no parity
Receives the commands controlling the module from the host, and sends the results
detected by the module and other information
＜USB＞
Outline
Communication System USB 2.0 *Using CDC class
Receives the commands controlling the module from the host, and sends the
detected by the module and other information
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
Specification
Specification
6
Page 7
White Paper
About the offering of the developing tool Specifications
■External Dimentions
＊”HVC” is an abbreviation of the Human Vision Components.
＊The name and logo of ”OKAOTM” are either registered trademarks or trademarks of OMRON Corporation
in Japan and/or other countries.
＊Windows is either a registered trademark or a trademark of Microsoft U.S. in the United States and/or other countries.
ELECTRONIC AND MECHANICAL COMPONENTS COMPANY
Image Sensor for Human Recognition Human Vision Components, “HVC-P2”
7
Page 8
White Paper
Please check each region's Terms & Conditions by region website.
OMRON Corporation
Electronic and Mechanical Components Company
Regional Contact
Americas
https://www.components.omron.com/
Asia-Pacific
https://ecb.omron.com.sg/
Korea
https://www.omron-ecb.co.kr/
© OMRON Corporation 2018 All Rights Reserved.
In the interest of product improvement, specifications are subject to change without notice.
Europe
http://components.omron.eu/
China
https://www.ecb.omron.com.cn/
Japan
https://www.omron.co.jp/ecb/
Cat. No. Y233-E1-01
0518 (0)
Loading...