细节展示放大镜特效

Face recognition

  • No:Face Recognition
  • Quantity:

    Race for intelligent portrait recognition product introduction

    Race for Intelligence in2008A special research and development team was set up to engage in the analysis and research of key technologies such as face detection, feature extraction and recognition algorithms in portrait recognition, and after many years of technical accumulation and innovation, the core algorithm with independent intellectual property rights was gradually formed, a number of provincial and municipal scientific research projects were undertaken, and a multi-position lighting constant human image access control system was developed.TI DSPThe platform's three product lines of high-performance portable portrait recognition products, as well as massive portrait retrieval systems.

    One. Race to be like a gated system

    Advanced recognition algorithm with independent intellectual property rights, using independently developed statistical model and light invariant technology first combined algorithm, the recognition speed is fast and the precision is high;

    With advanced intelligent self-learning algorithm, the database can be dynamically updated, eliminating the image differences caused by the years, and improving the recognition rate;

    Advanced image acquisition technology at large rotation angles (left and right)-35+35Degrees, up and down-20+20portraits collected under the scale can still be accurately identified;

    The combination of visible light and infrared completely eliminates the influence of light on recognition, and can be applied to various environments.

    The interface is simple and friendly, easy to operate;

    The network interface is rich, can be configured and managed through the mainstream browser;


    Key technical performance indicators:

    Database capacity

    Units (Sheets)

    Less than 100000

    Camera

    Size

    1/3"CCD

    Resolution

    640*480Pixel

    Image acquisition angle

    Around

    -35+35Degree

    Up and down

    -20+20Degree

    Working environment

    Power

    12V/1.52ASwitch power supply

    Operating temperature

    20+50

    Working humidity

    5%95%(Non-condensation)

    Appearance size

    Units (mm)

    148*165*43(Wide*Deep*High)

    Portable portrait recognition system

    The system is based onTI DSPPlatform, the portrait recognition algorithm for transplantation and optimization, equipment deployed in the airport, customs, stations, Ping An City, campus and other monitoring system front, through the camera captured images, identity verification.

    With advanced face detection technology, can quickly from the capture image accurately locate the position of the face;

    Not affected by race color, expression, tiara, eyes, beard, etc.

    The false positive rate is low, under the standard library conditions and library capacity, the false positive rate5%15%

    24Hourly non-stop work, with infrared recognition technology, night can still get a good recognition effect;

    Based onTI DSPPlatform, product performance is stable, environmental protection and energy saving;

    Interface introduction, friendly, built-inWebServer, can be managed through the mainstream browser configuration;

    Key technical indicators:


    Photo acquisition illumination

    Less than 10Lux

    Acquisition frame rate

    ³10Frame (black and white),³6Frame (Color)

    Portrait size

    More than horizontal pixels10%, greater than vertical pixels20%

    Agree to the maximum number of detected portraits in an image

    5

    Alarm response time

    Less than 1s

    Image library capacity

    Less than 100000(Zhang)


    Three. Race for a massive portrait retrieval system

    The main advanced portrait recognition algorithm and database management technology of the large library portrait retrieval system combine to search and contrast massive data in the multi-million-level or even multi-billion-level portrait database, thus providing an efficient means of finding. In the public security criminal investigation system, important places security, mobile population management and other aspects have an important application value.

    The system mainly consists of two parts, registration and identification. Registration is the building library section, which can be entered into images by file or capture, and multiple candidate results can be given by recognition similarity.


    Key performance indicators:

    Contrast speed

    Not less than10010,000 times

    Comparison accuracy of the library of tens of millions

    First hit is greater than85%Ago5Bit hit greater than90%

    Contrast speed

    Less than2s

    The minimum number of pixels for portraits

    48Pixel