HawkEye: AI-Powered Security Surveillance

🚀 Inspiration

Security incidents unfold in seconds, but traditional surveillance reacts too late. Security teams are forced to manually monitor multiple screens, leading to blind spots, delayed responses, and escalated conflicts.

We built HawkEye to change that. By combining AI-powered video analysis with real-time threat detection, we enable security teams to respond instantly—before incidents escalate. Our mission? Make surveillance proactive, not reactive.

🔍 What It Does

HawkEye is an AI-driven Anaomoly detection system that:
✅ Analyzes video footage using a deep learning model (ResNet50)
✅ Identifies violent altercations and classifies security threats
✅ Triggers instant alerts for security teams when a Anamoly is detected
✅ Displays real-time safety scores and incident reports via a web-based dashboard

With HawkEye, security teams can react before it's too late—enhancing public safety and minimizing risks.

🛠️ How We Built It

• Deep Learning Model: We trained ResNet50 on Anaomoly and non-anamoly datasets for high-accuracy detection.
  
• Backend: Developed with Flask, handling video uploads, frame extraction, and AI-based predictions.
  
• Frontend Dashboard: Built using HTML, Tailwind CSS, and Three.js for a sleek, futuristic UI.
  
• Alert System: Automated notifications ensure security teams receive immediate alerts upon threat detection.
  

🚧 Challenges We Ran Into

🔹 Training the Model: Achieving high precision in real-world fight detection required dataset improvements.
🔹 Reducing False Positives: Distinguishing aggressive behavior from harmless movements was challenging.
🔹 Optimizing Processing Speed: Maintaining low latency while analyzing multiple video frames in real-time.

🎯 What We Learned

• AI can revolutionize security monitoring, improving response times and situational awareness.
  
• Minimizing detection delay is critical—a few seconds can prevent serious incidents.
  
• User-friendly interfaces matter—security teams need a dashboard that’s intuitive and actionable in real time.
  

🚀 What’s Next

✅ Enhancing Model Accuracy with larger, more diverse datasets.
✅ Integrating Live CCTV Feeds for real-time, automated security monitoring.
✅ Expanding Alert Systems to integrate with emergency response teams (police, paramedics).
✅ Broadening Applications—adapting HawkEye for crowd management, law enforcement, and smart city security.

Built With

• pytorch
• tailwind