The concept revolves around leveraging artificial intelligence (AI) to analyze traffic camera footage to identify vehicles related to Amber Alerts. By utilizing the advanced capabilities of LLaMA, an AI model, the system can detect and classify vehicles based on their make, model, and color. This information is then cross-referenced with police alerts, providing real-time updates to law enforcement via SMS notifications, enabling rapid response and action.
The system operates through a series of structured processes. First, traffic cameras capture data by taking pictures at regular intervals, typically every five seconds. The captured images are then analyzed by the AI model, which has been specifically trained using LLaMA to identify vehicle characteristics such as make, model, and color. Once a vehicle matching the Amber Alert details is identified, the system compiles its findings into a detailed report that includes a picture, textual description, and precise location. This information is immediately sent to law enforcement via SMS, ensuring they are promptly notified with actionable insights.
The project began by sourcing data from a traffic camera in Pennsylvania, serving as a proof of concept. Images were captured from the live feed every two seconds. The LLaMA model was trained to recognize vehicle characteristics, and a program was created to match input details (e.g., make, model, and color) against the captured footage. The system loops through the available data, identifies potential matches, and sends the results, including location and details, to pre-specified SMS numbers for law enforcement use.
Building this system presented several challenges. One significant issue was sourcing suitable data. Initial datasets, such as Inirix's, captured images at five-minute intervals, introducing delays and inaccuracies. Additionally, the LLaMA model required highly specific formatting for image processing, necessitating time-consuming and meticulous formatting steps. Limited team size also posed obstacles, stretching resources thin. Errors during the deployment of the AI model further complicated progress. For the SMS notification system, AWS SNS services were initially considered but unavailable due to permission constraints. Alternatives required manual input of receiver phone numbers, service providers, and frequently updated passwords, making the process less seamless and more labor-intensive.
This system offers numerous advantages for various stakeholders. Law enforcement can respond to incidents more quickly and efficiently, improving public safety and increasing the likelihood of resolving crimes promptly. The public benefits from enhanced safety measures, fostering a greater sense of security. Cities can also allocate resources more effectively and achieve better crime resolution statistics, demonstrating improved governance and technology integration.
The future development of this system aims to address its current limitations. A primary goal is to secure faster and more reliable data sources to better accommodate urgent scenarios like Amber Alerts. Fine-tuning the LLaMA AI model will further enhance its detection accuracy and reliability. Additionally, the project seeks to implement a more robust and dependable SMS notification system to streamline communication with law enforcement, ensuring seamless functionality during critical situations.
Log in or sign up for Devpost to join the conversation.