SMART INTRUDER DETECTION USING IOT AND REAL-TIME OBJECT DETECTION

Abstract

ABSTRACT: Smart security systems increasingly use video surveillance. Due of their ability to promptly alert authorities if someone breaks the law or enters sensitive areas without permission. Deep learning and IoT are presented for a new intrusion detection system. Our proposed approach tracks the object's center of mass to determine entrance after applying the "You Only Look Once" (YOLO) method to find things. Remote siren systems can detect intruders, alert authorities, and enable response. SORT is also constantly monitoring this person. The NVIDIA Jetson TX2 programming tool builds and tests the real-time video transmission system. The average frame rate is 30, and it works 97% of the time. The intrusion detection system (IDS) should be adaptive so the user can pick between the reference frame that specifies the region of interest's dimensions and form and the data-free area. The system's pre-trained object classes can also identify people, cars, and animals as threats. Thus, it protects farms, cars, guests, and smart homes from animals in smart cities.