Inspiration
Current safety apps have a fundamental flaw: they are reactive. They require a user to unlock their phone, navigate to an app, and press a button during a moment of extreme trauma. In a real-world struggle, these seconds are the difference between safety and tragedy. Our inspiration was to create SafeHer AI—a proactive "Silent Guardian" that doesn't wait for a button press but instead uses AI to "listen" for distress and "feel" a struggle.
What it does
SafeHer AI is an advanced, multimodal safety hub that runs in the background to provide 360-degree protection:Audio AI (Voice Trigger): Constantly monitors for distress keywords like "Help Me" or "Bachao" and detects screams or crying frequencies.Motion AI (Shake Trigger): Uses accelerometer data to detect high-G force movements, automatically triggering an SOS if a physical struggle is detected.AI Risk Scoring: Generates a real-time risk index ($R$) based on environmental triggers and location data.Automated SOS Protocol: Instantly activates a 120dB siren, shares live GPS tracking, sends emergency SMS, and initiates automated calls to authorities.
How we built it
We built a full-stack solution designed for high-performance and low-latency:
Frontend: Developed using HTML5, CSS3, and JavaScript, featuring a futuristic Cyberpunk-themed dashboard with Glassmorphism effects.
Neural Core (Backend): Powered by Python Flask to handle real-time data processing and alert logging.
Sensing Layer: Integrated the Web Speech API for keyword detection and DeviceMotion API for struggle detection.
Deployment: Hosted on Render to provide a live, accessible URL for emergency testing.
Challenges we ran into
The journey was filled with technical hurdles:False Positive Calibration: It was difficult to distinguish between a user running/dancing and an actual struggle. We had to implement a threshold logic where the acceleration $a$ must exceed a specific limit $k$ ($a > k$) for a sustained period.Background Listening: Browsers have strict privacy policies. Ensuring the mic stays active and responsive without draining the battery required extensive optimization of the asynchronous event loop.UI/UX Balance: We had to ensure the UI looked professional and "Ultra-Tech" while remaining simple enough for a panicked user to understand at a glance.
Accomplishments that we're proud of
Successfully building a Hands-Free SOS system that triggers 100% via voice and motion.
Developing a Cyberpunk-style UI that stands out from typical flat-design safety apps.
Achieving a sub-1 second response time from trigger detection to siren activation.
What we learned
This project taught us the power of Multimodal AI. We learned how to synchronize different hardware sensors (microphone and accelerometer) to create a unified safety protocol. We also gained deep experience in full-stack Python development and the importance of user-centric design in high-stress scenarios.
What's next for SafeHer AI: Proactive Neural Emergency Response System
IoT Integration: Connecting with smartwatches and jewelry for haptic feedback alerts.
Edge AI: Implementing local machine learning models to detect distress even without an active internet connection.
Police PCR Linking: Integrating directly with city-wide emergency dispatch systems for faster response times.
Community Safety Heatmaps: Using anonymized data to create live "Safe-Zone" maps for all users.
Log in or sign up for Devpost to join the conversation.