W-Secure is a hybrid project that combines mobile app-based safety features with AI-powered CCTV surveillance to enhance women's safety in public and private spaces.
The W-Secure Android App is designed with discreet but powerful security tools to help women in distress:
- 🛰️ Live Location Sharing: Send real-time location updates to emergency contacts instantly.
- ☎️ Quick Helpline Calling: One-tap call to predefined emergency numbers.
- 🎥 Stealth Video Recording: Automatically records video in the background without opening the camera UI. The footage is stored on a secure cloud server to prevent tampering or deletion.
- 🗺️ Safe Space Mapping: View and mark nearby safe zones where help can be sought.
- 🆘 SOS Mode: Emergency trigger button that shares live location with trusted contacts and initiates other security protocols.
- 🔒 Shutdown Protection: Prevents unauthorized attempts to power off the phone during an emergency.
🔄 Update in Progress: We are currently working on making the location-based Safe Space detection more dynamic, with deeper analysis of surrounding areas using geofencing and real-time data.
The backend AI system is designed to integrate with CCTV feeds in public spaces and includes the following capabilities:
-
👥 Suspicious Following Detection:
- Tracks individuals and identifies situations where one woman is being followed by a group of men.
- Uses object detection and person tracking models to analyze motion patterns.
-
✋ Distress Gesture Recognition:
- Recognizes gestures such as raised hands or rapid arm movements that may indicate distress.
- Uses lightweight gesture recognition models based on MediaPipe or custom CNN-RNN architectures.
-
🚨 Alert Triggering:
- When a threat is detected, the system can trigger automated alerts to local authorities or monitoring teams.
- Platform: Android (Java/Kotlin/Flutter)
- Backend: Firebase (Realtime DB, Storage, Authentication)
- APIs: Google Maps, Geolocation Services
- Cloud: Firebase/Google Cloud for media storage
- YOLOv8: For real-time person detection.
- ResNet-50: For gender classification.
- MediaPipe: For gesture detection.
- OpenCV & DeepSORT: For multi-person tracking in CCTV feeds.
- Implement background video recording with stealth mode.
- Enable live SOS location sharing.
- Integrate Safe Space tagging on maps.
- Make Safe Space detection more dynamic using clustering and live crowd data.
- Deploy AI model on edge devices/CCTV systems.
- Improve gesture recognition accuracy.
- Create centralized admin dashboard for authorities.
We welcome contributions and collaborations from developers, researchers, and security experts. If you're passionate about using AI and tech for social good, feel free to open a pull request or reach out.
- 📧 Email: [hi@mayank4352.me]
- 🧑💻 Developer: [Mayank Sharma]
- 🔗 LinkedIn: [Mayank4352]
This project is built with the intention of helping people and enhancing safety. The AI-based surveillance is intended for authorized and ethical use only and must comply with local data privacy laws and regulations.