Challa Gomathi

AI Demos Website

Challa Gomathi

SentinelAI: Real-Time Security Camera Monitoring & Emergency Alert System

What the Project Does SentinelAI continuously ingests live or uploaded CCTV feeds via VideoDB, segments them into 30-second clips, and runs behavior- and object-analysis to detect crimes, assaults, or traffic accidents. It then automatically escalates high-risk events—alerting security teams or calling emergency services—to dramatically reduce human guard fatigue and response times. Core Functionalities & Data Flow Video Ingestion & Segmentation Connects to any IP or on-premise camera stream (or uploaded file). Uses VideoDB’s indexing API to slice into manageable 30-second segments. Semantic Analysis via VideoDB Sends each segment’s ID and a custom prompt to VideoDB’s “summarize video” endpoint. Retrieves rich metadata: people count, detected actions (e.g. fighting, loitering), object movements, environmental cues. Threat Assessment with LLM Feeds VideoDB’s JSON summary into a Groq-powered Llama model. Receives a structured threat profile: danger_level (low / medium / high / critical) reason (why it flagged) alert_required (boolean) Automated Alerting & Escalation Security Incidents: Instantly notifies on-site guards or remote monitoring teams via SMS, email, or dashboard push. Traffic Accidents: Triggers police/EMS dispatch protocols when a collision or injury is detected, supplying location and video evidence. Monitoring Dashboard Live video preview and segment playback. Real-time threat level indicators and confidence scores. Historical logs and downloadable JSON reports for audit and training. Resource Optimization & Guard Safety Reduces guard burnout by automating routine monitoring. Ensures rapid first response to violent crimes or emergencies, minimizing injury and property loss.

Yes

No

Yes

Yes

Yes, I’m interested