## Overview **AI Burglar Detector Alarm** is an open-source, real-time intruder detection system that turns your smartphone or laptop into a smart surveillance device — no CCTV required. Built using **COCO-SSD**, **TensorFlow.js**, and **React**, the system identifies unauthorized presence in front of your device camera and triggers an alarm on detection. 🚨 97% Drop in False Alarms • Peerlist Spotlight Top 1% • Zero Setup AI Security --- --- 97% Drop in False Alarms • Peerlist Spotlight Top 1% • Zero Setup AI Security --- AI Burglar Detector Alarm in action

--- ## Key Features - **Real-Time Intruder Detection**: Uses the **COCO-SSD** model via **TensorFlow.js** to detect people in the camera frame with high precision and low latency. - **Web-Based App**: No need to install anything. Just open the browser, grant camera access, and you're protected. - **Instant Alarm System**: The system raises a sound-based alarm the moment it detects a suspicious presence. - **Zero Cost Security**: No expensive CCTV setup. Just use any device with a webcam and internet access. - **Highly Customizable**: The open-source codebase allows users to tweak thresholds, customize alarms, and integrate with other smart systems. ## Technologies Used - **Next.js**: For server-side rendering and fast routing. - **React.js**: Component-based frontend structure. - **TensorFlow.js**: Real-time object detection with COCO-SSD pre-trained model. - **Tailwind CSS**: Utility-first CSS for responsive UI. - **React Webcam**: Captures camera feed directly in-browser. - **Lodash**: Utility functions for debouncing and performance optimization. ## Challenges and Learnings Building a real-time detection system in the browser without compromising performance was a major challenge. Tweaking the detection confidence and using **debouncing logic** from Lodash helped reduce false positives by **97%**. Cross-browser compatibility and ensuring privacy and security (without storing any camera feed) were also crucial to user trust and product credibility. Additionally, the project taught how edge AI can democratize security by reducing infrastructure costs and making protection accessible to everyone. ## Outcome - 🥷 **Turned any device into a burglar alarm** using just a web browser. - 📉 Achieved a **97% drop in false positives** by fine-tuning detection logic. - 🏆 Featured in **Peerlist Spotlight**, ranked Top 1% in trending tech projects. - 🌐 Live at: [ai-burglar-detection-ansh.vercel.app](https://ai-burglar-detection-ansh.vercel.app) --- ## Getting Started ### 1. Open the App Visit [ai-burglar-detection-ansh.vercel.app](https://ai-burglar-detection-ansh.vercel.app) in your web browser. No installation or signup is required. ### 2. Grant Camera Access When prompted, allow the app to access your device’s camera. This is essential for real-time detection. **Note:** No video is stored or sent to any server. ### 3. Position Your Device Place your laptop or phone so the camera covers the area you want to monitor. For best results, ensure the camera has a clear, unobstructed view. ### 4. Start Monitoring Once the app loads, it will automatically begin detecting people in the camera frame. If an intruder is detected, an alarm will sound instantly. --- ## Customization You can adjust detection sensitivity and alarm settings in the app’s settings panel: - **Detection Threshold:** Increase or decrease the confidence level for triggering the alarm. - **Alarm Sound:** Choose from different alarm sounds or mute the alarm. - **Notification Settings:** Enable browser notifications for silent alerts. --- ## Privacy & Security - **No Data Storage:** All processing happens locally in your browser. No images or video are uploaded or saved. - **Open Source:** Review or contribute to the code on [GitHub](https://github.com/AnshSinghSonkhia/AI-Burglar-Detection-website). --- ## Troubleshooting - **Camera Not Detected:** Ensure your browser has permission to access the camera. - **False Alarms:** Adjust the detection threshold in settings for your environment. - **No Alarm Sound:** Check your device volume and browser sound permissions. --- ## FAQ **Q: Does this work on mobile?** A: Yes, the app works on most modern smartphones and tablets with a browser and camera. **Q: Is my video feed private?** A: Yes, all detection runs locally. No video is sent or stored anywhere. **Q: Can I use this with external webcams?** A: Yes, select your preferred camera in your browser’s camera settings. --- For more help, visit the [GitHub Issues page](https://github.com/AnshSinghSonkhia/AI-Burglar-Detection-website/issues) or contact the project maintainer.