Abstract:
Closed Circuit Television (CCTV) systems are becoming more essential and being used in many official areas, such as public, private, social or working conditions, as well as houses for security purposes. CCTV cameras are installed in many cities and in many places of Pakistan, but in this project, we automated the camera by doing the task of human emotion detection, arm detection and identity hazardous situations by using deep learning. We used an algorithm that detect arm, human emotions and can alert the human when some arm or some dangerous situation is detected by CCTV camera. In this project human emotion detection and arm detection system is implemented using deep learning techniques. In our project human emotions are classified into six facial expressions Anger, Fear, Happy, Sad, Surprise and Neutral. We have also developed a mobile phone application which provides faster and very effective response time in detection of criminal activities by CCTV camera. To alert the concern person or to inform the nearby police station about the occurrence of crime, we have added an SMS sending module to our project application which sends SMS to concern person and to the police officers whenever such an event is detected. In this application we have also added a feature of sending video clip to nearby police stations to get their response as earlier as possible, in addition to a feature of calling to police when some dangerous situation is detected by our CCTV camera module. Proposed system in this work achieved the accuracy of detecting human emotions up to 64% and for object (Knives) detection to 65.7% which can be enhanced and improved in future extension of this work.
