Our Project
Inspiration
When hearing about the theft of The Four-Goat Square Zun, though the name is a bit odd, the look of the art piece is even stranger. We wanted to know who stole this precious artifact and why.
What it does
Our project does just that. We find out who stole the precious artifact
How we built it
We used AWS, Python, and Java to run AI and analysis of imagery taken from the scene. From here, we classify vehicles that match our targets.
Challenges we ran into
Our biggest challenge was finding accurate models to analyze the license plate. AWS Rekognition can only do so much with images, we decided we needed something more to be able to analyze both images and text together.
Accomplishments that we're proud of
We had over 80% accuracy in all 3 tests involving recognizing cars, with one of the challenges being 100%. Also, implementing OpenCV with the Amazon Rekognition helped improve the accuracy with the license plate recognition, reducing the number of false negatives.
What we learned
We learned a lot about AWS and ourselves in the process. This hackathon helped the group get to know each other better and work with AI tools that many of us were unfamiliar with. we also gained the opportunity to meet great members of both the AWS and MetroStar companies that helped sponsor this event.
What's next for Treasure Theft Hunt
We plan to continue working on our code to improve the accuracy of the license plate recognition software and create a presentation to share our methodologies and findings with others.
Log in or sign up for Devpost to join the conversation.