Face recognition and Alcohol Detection System

Executive Summary

Our customer is an early stage company and a leader in automotive security domain. Their solutions in automotive industries are well known for advanced features.

As per the traffic rules, drunk drivers are restricted to drive and to arrest them, the police has to work in a smart way. The customer had no expertise in image processing algorithms, so engaged eInfochips to develop a system to address this need to detect drunken vehicle drivers. Security is the major concern all over the world and many times faces have to be recognized for theft detection. The solution also required to recognize the drivers face with only 70% visibility which makes it useful in various theft detection cases.

With high level of expertise in image processing eInfochips developed the complete face recognition and alcohol detection system. Based on x86 platform we have developed this solution which can be used either independently or as a sub system of a bigger vehicle security system. We have enabled this solution to support Japanese language apart from English as our customer is a leading automotive security solution provider in Japan.

The Customer

Our customer provides industry’s most advanced automotive security solutions worldwide. Their solutions are well appreciated in various sectors including government.
With enhanced the features they are providing with automotive solutions for all size of security concern. They are developing a major security system for vehicle security management and eInfochips has developed this project as a subsystem enhancing customer’s major vehicle security management solution.

The Challenge

  • Enabling Face recognition and Alcohol detection system to recognize face in less visibility scenarios
  • To incorporate all the needed features for alcohol detection (for drunk driver test) and face recognition (for theft detection) in one single solution
  • To enhance system with multiple language support including Japanese language

The Solution

Design & Development
eInfochips leveraged its expertise on image processing and algorithm optimization for the development of Face recognition and Alcohol detection system based on x86 platforms and enabled it to support multiple languages.
To enhance face recognition system eInfochips enhanced system which allows 45 degrees of horizontal face movement in front of camera and to enable this feature we have developed the image enhancement algorithms.
The Face recognition Alcohol detection system needs to be used in night where there is less visibility and to support this requirement eInfochips developed Face Recognition algorithm which allows system and recognizing face in 70% visibility also.

face-recognition-system

We have incorporated the various security levels with Development of GUI required for enrollment and verification of the user as well as for testing of the application. Expert team of eInfochips developed various APIs for easy integration of Face Recognition System into the main system which enables APIs portability and easy interfacing with the main system.
To make solution more compact and comprehensive, team eInfochips formatted the image in desired forms such as JPEG, GIF, and TIFF according to the size of database.

How it works
The system consists of an alcohol detector with camera and an interlock main system, all patented by the customer. It has 3 operating modes:
Mode 1: Anti-theft test
The system captures the driver’s image for recognition of his face. If the driver is not recognized, it sends signal to the interlock main system to prevent the vehicle to start without an authorized driver.
After the successful recognition the system keeps capturing the driver’s image for verification at 1fps every 2 minutes.
Mode 2: Drunken Driving test
The system checks the alcohol level of the driver, and accordingly sends signals to the main system. If the alcohol level is above the specified safe limit, the system sends signal to the interlock main system preventing the vehicle to start with drunk driver.
After the successful recognition, the system keeps capturing live images every 2 minutes to verify the same driver, who was permitted to drive, at the initial stage of testing.
Mode 3: Anti-theft and Drunken Driving Test
This combines mode 1 and mode 2 to grant permission to the person to drive a vehicle.

The Technology

  • Board: X86 Platform
  • Language: C, MATLAB
  • Compiler/Tool chain: .Net, MATLAB
  • Development Environment: .Net, MATLAB

The Benefit

  • Team eInfochips reduced the project cost for customer over 20% with expertise in image processing
  • With multiple language support, eInfochips made the solution applicable worldwide and thus enabled the customer in targeting the global market
  • With Disciplined Execution as core value we developed and delivered project on prescheduled timings which helped customer in developing their major project related to vehicle security management system.