AuraHome is an IoT-based smart home security and safety platform that integrates access control, intrusion monitoring, hazard sensing, and real-time notifications within a single architecture. The system combines ESP32-based nodes with a Raspberry Pi 5 edge gateway for local processing, and a cloud layer that includes a Laravel backend, MySQL database, and a Mosquitto MQTT broker to enable secure publish–subscribe communication. A Flutter mobile application provides remote monitoring and control, while Firebase Cloud Messaging is used for near real-time alert delivery.Several AI components were integrated and evaluated independently. The Speech Emotion Recognition module achieved 77.45% accuracy on MSP-IMPROV (F1 0.775) and 31.59% on emoDB (F1 0.27), highlighting limited cross-lingual generalization. Gun detection, tested on 1,076 images, achieved precision 0.838, recall 0.835, mAP@0.5 0.872, and mAP@0.5:0.95 0.521. Masked-people detection, tested on 399 images, achieved precision 0.935, recall 0.740, mAP@0.5 0.856, and mAP@0.5:0.95 0.614. Face anti-spoofing was implemented using DeepFace without custom training and achieved 95.2% accuracy on around 150 diverse samples under varied lighting and spoofing attempts. Server-side load testing showed stable performance under moderate concurrency, while stress testing revealed capacity limits that motivate future scalability improvements. Preprint notice: This work is a preprint and has not been peer-reviewed.
Shaheen et al. (Thu,) studied this question.