Page 68 - Smart City Brochure
P. 68

Safety  and  Security  Automatic Facial Expression Recognition System (AFERS)
         Automatic Facial Expression Recognition System developed by CDAC, Kolkata can recognize six basic
         emotional  facial  expressions  of  human  beings  (irrespective  of  gender  and  age)  such  ashappiness,
         disgust, fear, surprise, anger and sadness from video sequences in online mode using camera followed
         by saving the video for offline analysis at a later time.
         Features:
                AFERS captures live video stream for online analysis and stores the video sequence for offline
                analysis.
                A unique feature of AFERS is quantifying with a score for each of the expressions based on the
                intensity or degree of expression exhibited by the person. The scores so exhibited have been
                validated by Human Psycho-visual judgment.
                The intensity of expression is represented graphically for one or more expression (in case of
                mixed expressions) for each of the frames.
                For easy user interpretation graphical representation of one or more expression is displayed
                after completion of analysis both in the form of bar graph and pie-chart.

















         Video Summarizer

         Video summarization and analysis comprise methods to create a summary of the video collection,
         identifying objects of interest, tracking of such objects of interests from frame to frame across multiple
         videos, and possibly identifying behavior or activities performed by the objects.
         The "Video Summarizer” assists and alerts the system operator about potential security threats.

         It provides the below features:
          Synopsis Generation: Short summarized surveillance video showing multiple activities occurred at
            different times, while preserving the essential activities of the original video.
          Indexing of Surveillance Video: pointing to the original video and the original time of activity

          Intrusion/Object Detection: Detection of moving objects in selected areas covered by the camera;
            Avoid false alarms due to wildlife or other moving objects (e.g., tree leaves). The object of interest
            could be tracked across several videos or it could simply be an object that has been left unattended in
            restricted areas covered by the cameras.

          Monitoring of Vehicles: Detection of illegal parking in no parking zones; detection of vehicles driving
            in wrong way or reversing.
          Crowd behavior:  Detection of people flow and counting of people in selected areas.
   63   64   65   66   67   68   69   70   71   72   73