My Hero

My Hero transforms any camera into a digital hero, detecting danger and silent calls for help the moment they happen and alerting authorities.

Comment 2

MyHero — AI-Powered Safety & Signal Detection

Inspiration

In a world where cameras are already installed almost everywhere, safety systems are still largely passive. Footage is often reviewed only after incidents occur, and real-time monitoring depends heavily on human attention, which doesn’t scale. Even more critically, many people in danger are unable to explicitly ask for help — whether due to fear, medical distress, coercion, or physical limitations.

We were inspired by real-world situations where subtle or secret distress signals went unnoticed: silent hand gestures, unusual body posture, or behavior indicating danger that didn’t look overtly violent. These moments highlighted a major gap in existing surveillance solutions. We wanted to build a system that doesn’t just record events, but understands context, intent, and human behavior, and can act immediately. MyHero was created to transform ordinary cameras into proactive safety heroes that respond when people need help most — even when they can’t say it out loud.

What it does

MyHero is an intelligent, real-time video analysis platform that detects dangerous situations, criminal activity, medical emergencies, and covert distress signals using live camera feeds or uploaded video. It can identify events such as violence, fainting, suspicious behavior, and predefined secret gestures, then instantly alert security or responders through automated notifications.

The platform generates time-stamped incident reports with visual evidence, maintains a searchable library of analyzed footage, and provides an intuitive dashboard for monitoring multiple cameras simultaneously. MyHero also includes an AI-powered assistant that understands the ongoing situation and can answer contextual questions like “What is happening right now?” or “What should I do?”, helping users make fast, informed decisions during critical moments.

we also implemented a hardware response layer by connecting MyHero to an Arduino-controlled buzzer. When a threat or predefined distress signal is detected, the system can trigger an immediate physical alert, demonstrating how MyHero bridges AI-driven video analysis with real-world safety responses.

How we built it

MyHero was built using a modern, scalable stack optimized for real-time performance and AI-driven analysis:

  • Frontend: Built with Next.js and TypeScript, delivering a fast, responsive interface for live video feeds, event timelines, statistics, and analytics.
  • Backend: A Python-based backend handles video ingestion, event processing, authentication, alerting logic, and data storage. Python was chosen for its powerful ecosystem in AI, computer vision, and real-time processing.
  • AI & Computer Vision: Live audio and video streams are analyzed using pose estimation, gesture recognition, and multimodal vision-language models to detect visible threats, medical emergencies, and secret distress signals.
  • Real-Time Communication: WebSocket-based pipelines enable low-latency updates, ensuring detections and alerts are delivered instantly.
  • Alerting & Assistance: Automated email and phone notifications are triggered when incidents occur, and an AI assistant provides context-aware guidance based on real-time event data.

The system is accessible through any modern browser while remaining scalable enough to support multiple cameras and environments.

Challenges we ran into

  • Performance Optimization: Balancing real-time video processing with browser performance and AI model rate limits.
  • AI Model Accuracy: Fine-tuning detection algorithms to minimize false positives, especially for subtle gestures and edge cases.
  • Video Stream Handling: Managing multiple simultaneous video streams without overwhelming the system or degrading user experience.

Accomplishments that we're proud of

  • Created a fully functional AI-powered surveillance system in 24 hours
  • Achieved real-time processing with minimal latency
  • Implemented a beautiful, intuitive user interface
  • Built a scalable architecture capable of handling multiple cameras
  • Developed a system that is accessible through any modern browser -Integrated a hardware response layer using an Arduino-controlled buzzer to trigger immediate physical alerts upon threat or predefined signal detection.

What we learned

  • Advanced video processing techniques directly in the browser
  • Real-time data handling using WebSocket connections for live updates
  • AI model optimization for edge cases and ambiguous scenarios
  • Complex state management in React applications dealing with large datasets
  • Integration of multiple third-party services into a cohesive system
  • The critical importance of user experience in safety and security applications

What's next for MyHero

Future enhancements we’re planning include:

1. Advanced AI Features

  • Person identification and recognition
  • Object tracking across multiple cameras
  • Behavioral pattern analysis

2. Enhanced Security

  • Advanced access control and permissions

3. Smart Environment Integration

  • Integration with popular smart home and building platforms
  • Automated response actions
  • Voice assistant compatibility

Our vision is to make MyHero the go-to platform for intelligent, proactive safety monitoring — helping organizations and communities respond faster, smarter, and more effectively when it matters most.

Built With

Python
Next.js
TypeScript
Computer Vision
Vision-Language Models
WebSockets
AI & Machine Learning

Built With

Share this project:

Updates