CopVision

Inspiration

Our team thought a lot about opportunities that we could do involving the use of a camera, as it was something we were all interested in learning more about, and after discussing our experiences of extensive handwritten paperwork, we were excited to create an AI model that could help this process. When we came across the body camera idea, it felt like a no-brainer combination of the ideas we had been most interested in while brainstorming.

What it does

Our AI-powered model helps expedite the lengthy process of law enforcement documentation and offers a significantly easier way for police officers to analyze a case based on their body camera footage.

How we built it

The backend model was developed with Whisper AI, a machine learning model to help transcribe and summarize the audio from a body cam recording. We also used OpenAI in cohesion with Blip-2 to help with analysis of the video and tracking movements. The rest of the backend communication was written in Python. The front end was developed using a React.js framework, along with JavaScript and CSS for the interface. We also implemented React-Three to create a three-dimensional model to help with our data visualization.

Challenges we ran into

It was our first time approaching video analysis, so it was very challenging to figure out what existing software was available to help and what steps we would need to take for our specific use case and dataset. Additionally, integrating multiple AI models into a seamless pipeline required a deep understanding of API calls, data processing, and performance optimization.

Accomplishments that we're proud of

We are proud of successfully integrating multiple AI models to create a working prototype that automates law enforcement documentation. Our team worked hard to overcome the learning curve of video and audio processing, and we were able to create a functional system that processes body camera footage efficiently. Another major achievement was implementing an intuitive user interface that allows officers to interact with the data in a meaningful way.

What we learned

Throughout this project, we gained valuable experience in AI model integration, video processing, and working with real-time data. We also deepened our knowledge of front-end development, particularly in React and React-Three, and learned how to optimize performance when handling large-scale video and audio inputs. Additionally, we developed a greater appreciation for the real-world challenges of law enforcement documentation and how AI can be used to improve efficiency and accuracy in these workflows.

What's next for CopVision

Moving forward, we plan to refine our AI models to improve accuracy and efficiency in transcription and analysis. We also want to explore real-time processing capabilities to provide instant insights during active incidents. Additionally, we hope to implement a secure cloud-based storage system to ensure accessibility and compliance with legal and privacy regulations. Another goal is to collaborate with law enforcement agencies to receive real-world feedback and tailor the tool to better meet their needs.

Built With

• css
• javascript
• python
• react
• whisper