Rapidresq: AI-Based Traffic Accident Detection and IoT-Enabled Emergency Notification Grid

Abstract

Road traffic accidents continue to be a major global public safety concern, especially in areas with inadequate communications infrastructure. Conventional surveillance relies on manual observation, which causes response delays and human fatigue. RapidResQ, an autonomous real-time collision detection system utilizing edge computing on the ESP32-CAM platform, is presented in this paper. To minimize latency and eliminate the need for GPU-class edge hardware, cloud-assisted inference is made possible by a quantized Fast-CNN derived from MobileNetV2 that is deployed via Google Teachable Machine and accessed over local Wi-Fi.. False positives are decreased from 18.3% to 6.8% using a two-tier validation mechanism that combines AI visual classification with analog sensor threshold verification. A hybrid architecture is used for emergency alerting: GSM-enabled SMS with voice telephony for rural areas and IoT cloud messaging for urban areas. With an alert latency of less than five seconds, evaluation attains 90.5% classification accuracy.