SnapReport

Web Application
User Input
output
Admin Portal

What Motivated Us

According to National Emergency Number Association, roughly 240 million Americans call 911 each year in moments of distress. While emergency services aim to respond quickly, the reality is that for someone in an urgent situation, every minute of waiting feels like an eternity.

As shown in the table below, the time it takes for first responders to arrive can range from under 5 minutes to over an hour, depending on the situation. For instance, 32.3% of robberies receive a response within 5 minutes, but over 28% take 11 minutes to an hour. Aggravated assaults see similar delays, with 36.4% taking up to an hour, and 5.4% having unknown response times. For property crimes, nearly half (47.8%) take 11 minutes to an hour, and 12.6% take over a day.

Type of Crime	Within 5 min	6-10 min	11 min -1 hr	Within a Day	1+ Day	Unknown
Violence	28.30%	30.30%	33.50%	2.50%	0.40%	5.00%
Robbery	32.30%	38.80%	28.30%	0.00%	0.00%	0.60%
Aggravated Assult	20.90%	32.60%	36.40%	4.80%	0.00%	5.40%
Simple Assult	31.40%	28.00%	31.80%	2.40%	0.10%	6.20%
Property Crimes	12.80%	20.20%	47.80%	12.60%	1.90%	4.70%
Household Burglary	13.60%	21.80%	46.90%	12.60%	1.90%	3.30%
Motor Vehicle Theft	12.50%	22.20%	49.10%	11.50%	1.30%	3.30%
Theft	12.50%	18.90%	48.00%	12.70%	2.00%	5.90%

Data sourced from Safe Smart Living

These numbers highlight a crucial issue: every minute matters in life-threatening emergencies. Our project uses AI to reduce response times by delivering immediate insights, helping first responders prioritize more effectively. By speeding up analysis and action, we aim to improve outcomes and get help to those in need faster.

What SnapReport Does

SnapReport is an AI-powered platform that empowers users to quickly report emergencies by uploading images, videos, and descriptions. SnapReport analyzes this information in real-time to assess the situation and determine the necessary emergency response, whether it's police, firefighters, or paramedics.

By seamlessly combining visual data, user input, and precise location tracking, SnapReport provides first responders with critical insights, ensuring faster and more efficient action. This approach helps to reduce response times and enhance safety for everyone involved.

The Technology Behind It

SnapReport is a mobile and web application built with a blend of technologies. The frontend is developed using React Native and ReactJS, while the backend uses Flask and implements the Model-View-Controller architecture, seamlessly integrated with our AI models. We developed our initial model using Google Colab and deployed our backend on Google Cloud.

For AI processing, we utilize the OpenAI API, which handles initial image analysis and generates detailed descriptions of emergencies. These are then processed by the GPT-4 Turbo model to ensure structured information, comparison with user inputs, and tailored response recommendations. Additionally, we use the Google Maps API to post geo coordinates on our admin page for precise location tracking.

To streamline data storage and ensure continuity, SnapReport integrates Neurelo's API, securely managing insights and analysis without the complexities of database programming. We also use MongoDB to efficiently handle unstructured data in JSON format, enhancing overall efficiency.

Overcoming the Obstacles

A major hurdle was getting SnapReport’s AI to accurately process and analyze emergency images in real-time, as each situation is unique and complex. We overcame this by refining the models and applying advanced techniques to improve recognition and data extraction.

Another challenge was bridging the gap between AI analysis and user input. To resolve this, we built a system to detect and flag discrepancies, ensuring users receive clear, reliable recommendations.

Takeaways

Our first hackathon experience led us to create SnapReport, a tool aimed at speeding up emergency response times. Inspired by alarming statistics about 911 call wait times, we developed an AI-powered platform that analyzes uploaded images and descriptions of emergencies to quickly determine the appropriate response.

Building SnapReport was challenging, as we had to rapidly learn new technologies and overcome obstacles in real-time AI analysis. Despite the steep learning curve, we're proud of what we've achieved. SnapReport not only represents our growth as developers but also has the potential to make a real difference in emergency situations, potentially saving lives by providing first responders with crucial information faster.

The Road Ahead

As we look to improve SnapReport, we've identified a few key areas for future development. We're planning to expand our AI's capabilities to handle a wider range of emergency scenarios, which should make our system more versatile. We're also considering adding Voice-to-Text technology and video upload features to give users more ways to report emergencies.

One limitation we're aware of is the potential for false reports. To address this, we're exploring the idea of implementing a fake request detection system, possibly using a combination of AI and human verification. These enhancements should help make SnapReport more reliable, ultimately improving its effectiveness as an emergency response tool.