Saheli - Your Ultimate Safety Companion

Inspiration

The inspiration for Saheli came from a deep-rooted desire to address the pressing issue of women's safety in everyday life. Growing up in an environment where safety was a constant concern for women, We were motivated to create a solution that could empower women to confidently navigate through various situations. During my interaction with women from diverse backgrounds , we realized the significance of using technology to bridge safety gaps. Thus, Saheli was born, designed to not only protect but also give women the tools to take control of their safety.

What it does

Saheli is a comprehensive safety application aimed at ensuring women's security through innovative features:

• SOS Alert: Sends real-time alerts with location details to trusted contacts during emergencies.
• Safe Route: Provides the safest and shortest route using Google Maps API and K-means clustering.
• Violence Detection with Smart Glasses: Detects violent behavior using ML models and triggers SOS alerts if no response is received.
• Accident Prevention: Detects harsh brakes while riding and alerts the user to prevent accidents.
• Emergency Services Locator: Locates nearby police stations, hospitals, and bus stands for quick access.
• Helpline Directory: Lists essential helpline numbers for easy reference.
• Fake Caller: Discreetly triggers fake calls to help users escape unsafe situations.
• AI Chatbot (Sakha): Provides safety tips, guidance, and emergency assistance through a chatbot powered by Gemini AI.
• Crowdsourced Safety Zones: Allows users to report unsafe areas and identify safe zones, fostering a community-driven approach.
• Offline Accessibility: Ensures functionality even without an internet connection.
• Audio-Video Recording: Automatically records and streams audio and video during emergencies for evidence.

How we built it

We built Saheli using a combination of technologies:

• Frontend: Flutter & Dart were used to create an intuitive, cross-platform user interface.
• Backend: Node.js and Express.js powered the backend, with Firebase for authentication and Firestore for the database.
• Machine Learning: For violence detection, we used TensorFlow and OpenCV to build a model capable of analyzing video streams for violent behavior.
• Smart Glass Integration: Raspberry Pi was used to integrate smart glasses for real-time video streaming and emergency notifications.
• AI Chatbot (Sakha): We leveraged Gemini AI, Vertex AI API, and Dialogflow to create a conversational assistant capable of providing safety guidance and responding to emergencies.
• Deployment: The app was deployed on Google Cloud, ensuring scalability and performance.

Challenges we ran into

• Integration with Smart Glasses: Integrating real-time video streaming from Raspberry Pi and syncing it with the app posed a challenge. However, after much trial and error, we succeeded in establishing a reliable connection between the glasses and the backend.
• Machine Learning for Violence Detection: Training an accurate machine learning model for violence detection was challenging due to limited training data. We had to carefully curate datasets and refine the model using TensorFlow to ensure reliability.
• Flutter and Backend Communication: Ensuring smooth communication between the Flutter app and the Node.js backend was initially tricky due to the asynchronous nature of both. Optimizing API calls for better response times took several iterations.
• Offline Functionality: Ensuring that core features worked seamlessly offline required careful planning and use of local storage to cache vital data without compromising performance.

Accomplishments that we're proud of

• Impactful Features: The integration of Violence Detection, AI Chatbot, and Crowdsourced Safety Zones has proven to be a groundbreaking step towards addressing real-world safety concerns.
• Community Involvement: The crowdsourced approach to reporting unsafe areas and creating safe zones is empowering women to actively participate in building a safer environment for all.
• Global Reach: Saheli has garnered attention from global tech communities and is helping women across the world feel more secure.

What we learned

• Importance of User-Centric Design: The project reinforced the importance of designing an application with the user at the center. We constantly tested and iterated on the app based on user feedback, which helped us improve usability and functionality.
• Collaboration and Teamwork: This project highlighted the power of collaboration. Working with a talented team of engineers, designers, and safety experts taught me the value of bringing diverse skill sets together to solve complex problems.
• Tech for Good: We realized that technology can be a powerful enabler for social change. By combining innovation with a real-world purpose, we have the ability to create impactful solutions that improve lives.

What's next for Saheli - Your Ultimate Safety Companion

• Expanded Partnerships: We plan to partner with local law enforcement, government bodies, and NGOs to enhance our features and data sources. By incorporating crime data and street lighting information, we can make the Safe Route feature even more accurate.
• Real-Time Safety Alerts: We aim to implement real-time safety alerts based on user location, integrating with existing crime and incident reporting systems.
• Gamification and Rewards: We will introduce gamification elements to encourage users to contribute to the crowdsourcing of safety zones and reporting unsafe areas, rewarding active participants.
• AI-Driven Insights: By further improving the AI capabilities, we aim to provide personalized safety insights based on the user’s location, behavior, and preferences.
• Global Outreach: Our mission is to expand Saheli’s reach to more countries and regions, ensuring that every woman, no matter where she is, has the tools and resources to stay safe.

We believe Saheli is just the beginning of a larger movement that empowers women to live freely and fearlessly.

Built With

• apis:-google-maps
• databases:-firebase-firestore
• dialogflow
• firebase
• frameworks-&-libraries:-flutter
• javascript
• k-means-clustering
• languages:-dart
• opencv
• other-technologies:-gemini-ai
• platforms-&-cloud-services:-google-cloud
• python
• raspberry-pi
• tensorflow
• vertex-ai