Kuronami

This project offers AI models for fire/smoke, road accidents, and violence detection with Twilio alerts. Easily deploy via a web interface for real-time safety monitoring and rapid emergency response.

Comment

About the Project

This project is an AI-powered safety monitoring system designed to detect fire/smoke, road accidents, and acts of violence, all integrated into a user-friendly website. The idea was born from the urgent need for real-time alert systems in environments where quick responses can save lives. By harnessing the power of machine learning and automated notifications, we aimed to build a reliable system for enhancing safety and reducing emergency response times.

Inspiration

The project was inspired by the increasing number of preventable tragedies that occur due to delays in detecting and responding to emergencies. Whether it's a fire breaking out, a road accident happening in an isolated area, or violence in public spaces, early detection is crucial. The possibility of combining AI's capabilities with real-time alert systems inspired us to create a solution that could potentially save lives.

What it does

This system leverages AI models to monitor video feeds and detect fire/smoke, road accidents, and violent activities. Upon detecting an incident, it sends real-time alerts via SMS using Twilio, allowing rapid response and intervention. The accompanying website provides a user-friendly interface for monitoring and reviewing incidents.

How we built it

Model Selection and Training:

Developed three individual models for fire/smoke detection, road accidents, and violence detection using PyTorch and OpenCV.

Trained the models on specialized datasets to achieve high accuracy.

Alert System Integration:

Integrated Twilio for sending real-time alerts to predefined recipients.

Web Deployment:

Created a simple yet functional website using Flask to host the models.

Integrated the models into the website and ensured smooth communication between the front end and back end.

Testing and Refinement:

Tested the system with sample videos to ensure reliable performance.

Made iterative improvements based on detected issues and user feedback.

Challenges we ran into

Model Path Configuration: Ensuring that the paths for the models and sample videos were correctly set up on different systems.

Performance Optimization: Balancing accuracy and speed for real-time video processing.

Alert Customization: Configuring Twilio for seamless and secure alerting.

File Size Limitations: Managing large files and ensuring the web application could handle them efficiently.

Integration Issues: Merging three independent models into a cohesive system that functions smoothly within the website framework.

Accomplishments that we're proud of

Successfully integrating three AI models into a single web application.

Achieving high accuracy in detecting incidents across diverse scenarios.

Implementing a real-time alert system to enhance safety.

Building a user-friendly interface for monitoring and reviewing detected incidents.

What we learned

Throughout this project, we gained a deeper understanding of:

Computer Vision Techniques: Implementing object detection and activity recognition to identify fire, accidents, and violent behavior in videos.

Model Optimization: Fine-tuning models for better accuracy and faster performance.

Alert Integration: Utilizing Twilio for real-time SMS alerts to notify users of detected incidents.

Web Development: Building and deploying a local website to provide a user interface for monitoring.

Problem-Solving: Tackling issues like model path configuration, video format compatibility, and efficient file handling.

What's next for Kuronami

Expanded Detection Capabilities: Incorporating additional models for broader incident detection, such as theft or medical emergencies.

Cloud Deployment: Moving the system to the cloud for enhanced scalability and accessibility.

Mobile App Integration: Developing a mobile app to complement the web interface and offer on-the-go monitoring.

Improved Alert System: Adding more customization options for alerts, including email and push notifications.

Real-Time Data Analytics: Providing detailed analytics on detected incidents for better insights and preventive measures.

Built With

Share this project:

Updates