Intelligent Surveillance System Leveraging IoT for Enhanced Situational Awareness

Abstract

Home surveillance systems are still challenging, particularly for patrolling or tracking subjects through CCTV images despite recent developments. Therefore, it is crucial to instantly identify human intruders in real time based on motion and face recognition. The proposed model represents a cost-efficient real-time Intelligent surveillance system for home and small offices using raspberry pi, PIR Sensor, and computer vision. The proposed system detects for motion with PIR sensor in its field of view and starts video capturing once motion is detected and sends it to user via email. HOG Descriptor and SVM Classifier is used to differentiate between human and inhuman objects. Haarcascade filter is implemented to detect intruders jumping over the wall. For Facial recognition, Images are captured or imported to create a database with help of frontal face LBHP filter. This database is used to train a facial recognition model. The trained facial recognition model can recognize authorized and unauthorized person. In our proposed model, the system tracks the detected individuals face in the frame and only focuses on the image content in these facial regions. Then, LBHP filter is used for recognizing detected faces based on the pre-provided face database and differentiate as known or unknown user. The system works satisfactorily in normal lighting conditions with accepted accuracy.