MegaHack is a 24-hour hackathon held at Virginia Commonwealth University (VCU). It is a sprint of intense innovation and problem-solving for developers and makers to meet challenges. This year, as a proud sponsor, CCI CVN set the challenge in categories based on our research areas of focus — Smart Health and Smart Cities themes attracted numerous student teams, who also received ongoing support from the Node’s mentors throughout the event.
The “Best Smart City Hack” award went to Advay Choudhury and Nand Vinchhi for their Realert project.
Realert aims to improve emergency response times and safety in schools by utilizing machine learning threat detection, fed with a combination of live data from devices that are already in schools like mobile phones, and existing CCTV cameras.
“In true hackathon fashion, they looked at how to utilize open-source datasets and technology to support their ambitious 24-hour project goals. A mix of object tracking, video processing, and audio analysis, this project has to consider the trade off between privacy, data collection methods, emergency necessity, and ease-of-use in a high-stress situation, which are all important factors in a city environment where technology is becoming increasingly pervasive,” says Lauren Linkous – a Ph.D. candidate, mentor at Megahack, and a principal figure in our MDS lab.
Key features:
- Location tracking: Uses mobile app data and CCTV camera information to accurately locate individuals within the school building.
“Seeing our object detection model successfully identify a simulated threat in the CCTV footage was a key moment. It proved the functionality of our system and its potential utility for emergency response,” said Advay. - Gun violence detection: Employs a machine learning model for audio and video analysis to detect potential gunshots.
“Integrating our audio classification model into the app was a major challenge. We worked through the night, debugging and testing until we finally got it to work. It was a crucial step forward for our project,” said Nand.
- Responsive alerts: Real-time alerts with accurate location information are sent to students, staff, and emergency responders via mobile app and SMS.
Screenshot of a live alert from the Nand and Advay’s demo presentation.
Challenges:
- Data integration: Seamlessly integrating data from various sources like phones and cameras in real-time.
- Accuracy: Ensuring the reliability of both location tracking and gun violence detection models.
- Privacy concerns: Balancing security needs with individual privacy considerations.
Potential benefits:
- Faster response times: Enables quicker intervention by authorities in emergency situations.
- Improved situational awareness: Provides real-time information to individuals and responders about threats and locations.
- Deterrence: Potential deterrent effect on criminal activities within the school environment.
Overall, Realert presents a promising approach to enhancing emergency response and safety in educational institutions. While facing challenges in data integration, accuracy, and privacy, it could significantly improve safety outcomes for students, staff, and the community.
If you would like to learn more about Realert, you can visit the team’s website or read their article on Devpost.
