Computer vision or image processing, is an important field of modern day computer science. Computer vision was a subject of research since 60’s, but in recent days it is being used to solve real-life problems and producing excellent results. Computer vision, along with deep learning can be used to develop facial recognition systems, which is vastly used in the fields like criminal detection, crowd monitoring, attendance systems etc.
We have all come across systems based on biometrics and RFID cards. But compared to theses dated technologies, the approach of direct and reliable Face Recognition opens smarter and better solutions. Algonics has developed cutting edge expertise on Face recognition technology. With the help of such technology, we can detect the presence of known persons across an establishment.
Algonics have developed an attendance and automatic people recognition system (we have named
it Smart Log
) which is
able to recognize known and unknown people in an establishment. The system needs to be
trained
with some photographs of a person. Thenceforth, the system automatically
detects trained persons with a high degree of accuracy, whenever the person comes in
the field of appropriate view of the cameras. We can deploy this system in an existing
networked CCTV setup, and it also supports Android smartphones costing a few thousand
rupees.
We can apply such system for different smart solutions including:
CASE 1: Detection and Categorisation of all visitors.
The technology of face recognition is appropriate to counter incidents of unwanted
persons visiting an establishment. With a camera placed in a strategic place, we
can
categorise all visitors in three groups, White list, Black list
and Neutral list. The end-user shall get a ready data of all persons coming under
the
view of
one or more cameras categorised as per the above list along with time and date
stamp.
CASE 2: Detection of persons in multiple zones:
All establishments have certain restricted zones where entry is subject to suitable
authorisation and permissions. Movement of persons in such zones can be
automatically
tracked and alarms raised accordingly by use of camera based face recognition
systems.
Suitable reports may be generated with critical matches along time lines.
CASE 3: Detection of persons across multiple geographical areas
When a system is configured
with automatic face recognition and connected to the internet, we can provide
statistics on the time required for different persons to reach
from one office to another office or business place. We can also match the white
listing, black listing and neutral listing
across different premises located in distant places.
Watch the video below to get a clear idea about our system :
The accuracy of the face recognition systems depends on multiple parameters which includes pose, illumination, facial expression, age, and occlusion. Real-time face detection and tracking in the normal indoor environment is well solved and we have reached 90% to 95% of accuracy depending on the algorithm used. The speed of the system depends entirely on the hardware capabilities of the platform and the algorithm used. Tuning the hardware along with the algorithm, we can process upto 10 frames per second on a single computer based system.
For more articles Click Here , or to view our projects visit showcase.