Defenshe | Devpost

Idea
Features
Web
App

Inspiration

Women’s safety remains a major concern, with traditional solutions being reactive rather than preventive. Existing safety apps rely on SOS alerts after an incident begins, often leading to delayed responses. Inspired by the need for real-time prevention, we developed Defenshe—an AI-driven safety ecosystem that proactively detects threats and ensures immediate assistance.

What it does

Defenshe is a multi-layered safety solution that integrates: Web Dashboard: Law enforcement and authorities monitor live threats via CCTV-integrated AI. Mobile App: Users can send SOS alerts, share locations, detect hotspots, and access emergency contacts. Smart Wearable: Features include a panic creator emitting high-frequency alerts to deter attackers, accelerometer & gyroscope for detecting distress, and a hidden SOS button.

How I built it

Web & App: Built using React, Flutter, Firebase, and Node.js for seamless user experience. AI Models: Integrated YOLOv8 & custom ML models for gender detection, crowd analysis, and hotspot detection. Hardware: Developed a compact wearable with** ESP32, accelerometer, gyroscope, distance sensor, and a panic creator.**

Challenges I ran into

Fine-tuning ML models for real-time detection. Ensuring low latency communication between the wearable and the app. Optimizing the panic creator’s frequency range to balance deterrence and user safety.

Accomplishments that I'm proud of

Won first prize on a national level hackathon with this software which had over 2000+ registerations.

What I learned

Optimizing AI models for real-time threat detection. Hardware-software integration for seamless safety features. The importance of user-centric design in life-saving applications.

What's next for Defenshe

Expanding AI models to detect weaponry and unusual activity. Enhancing wearable features with GPS-independent tracking **for remote areas. Partnering with **governments & universities for large-scale implementation. Introducing community-based alerts for crowd-sourced safety reporting.