SentinelVision

Inspiration

The increasing threat of armed robbery with the alarming United States statistics, which recorded 44,086 incidents involving handguns in 2022 alone inspired us to develop a solution that enhances safety for store owners and employees. With the rise of technology, we envisioned a system; SentinelVision which aims to mitigate these risks by leveraging advanced AI technology for rapid gun detection and immediate response.

What it does

SentinelVision is an innovative gun detection system that processes uploaded video feeds in real-time. Upon detecting a firearm, the system not only alerts authorities immediately but also saves the frames where the gun is identified. These frames are securely transmitted to law enforcement, aiding in suspect identification and apprehension.

How we built it

We leveraged the State of the art recently released YOLOv10 small model, trained on a dataset of 9,256 annotated images, to achieve high accuracy in gun detection. The application preprocesses video frames using techniques like CLAHE for enhanced contrast and gamma correction for brightness adjustment. Frames with detected guns are saved locally and sent to police servers for further analysis.

Challenges we ran into

Some of the challenges included:

Acquiring a diverse and comprehensive dataset to ensure accurate model training.
Implementing seamless integration for real-time video processing and frame saving.
Optimizing the application's performance to handle high-resolution video streams without compromising speed.

Accomplishments that we're proud of

We are proud of:

Successfully developing a robust gun detection system capable of real-time operation.
Establishing a secure framework for saving and transmitting sensitive data to law enforcement agencies.
Creating an intuitive user interface that enhances usability for business owners and law enforcement personnel.

What we learned

Through this project, we gained valuable insights into advanced AI model deployment, real-time video processing challenges, and the complexities of integrating security solutions into existing infrastructures. We also deepened our understanding of legal and ethical considerations surrounding data sharing in security applications.

What's next for SentinelVision

Future plans include:

Enhancing the model to detect and classify other threats and suspicious activities.
Partnering with law enforcement agencies to optimize incident response protocols.
Introducing additional features such as facial recognition for enhanced suspect identification.