Umudike Journal of Engineering and Technology

The increasing demand for reliable and intelligent security systems has driven the integration of artificial intelligence with traditional sensor-based intrusion detection methods. This paper presents the design and implementation of an advanced real-time smart security system that combines Passive Infrared (PIR) sensing with deep learning-based object detection to enhance detection accuracy and reduce false alarms. Upon motion detection, a Raspberry Pi–based camera module captures images which are processed using a Single Shot MultiBox Detector (SSD) with a MobileNetV3 backbone, enabling real-time identification and classification of human intruders. The proposed system incorporates multiple alert mechanisms, including a buzzer, voice-based speaker alerts, and mobile push notifications containing image and text data. Experimental evaluation demonstrates that properly calibrated PIR sensors and adequate illumination reduce false positives by 35% and achieving a mean object detection confidence threshold of 76.2%. Notification latency analysis shows rapid response for local alerts (0.1-0.5 s), while richer push notifications incur higher delays due to network and processing overhead. The results validate the effectiveness of the hybrid sensing and deep learning approach in improving detection reliability, responsiveness, and practical applicability for smart security and surveillance applications.