Team 22 - AceGlass (Hardware)

Inspiration

We've seen people use Meta glasses and other AR devices to do everything from recognize faces on walks to play chess, and we thought: 'Why not make our own glasses?' So, we did! We were also inspired by our interest in game theory and how it applied to the game of poker. Beyond psychology, we wanted to discover live answers to questions like: 'When do I know to bet, and how much should I bet, empirically?' It's pretty hard for new players to memorize more than 7,000 possible hand combinations in poker, so we thought this would be a great start.

What it does

Put on our glasses and headphones and be transported into a world of poker mastery. By using a camera connected to a Raspberry Pi 5, Ace interprets its surroundings and finds playing cards. It classifies player's cards and dealer cards, and uses an API call to determine hand strength relative to the board, and provides a play recommendation straight to your ear.

How we built it

Using an RPI camera plugged into a Raspberry Pi 5, we fed the images from the glasses through an OpenCV algorithm to detect the contours of cards, classify them by rank and suit, and assign them either as hand cards or table cards. From there, we evaluated the hand strength locally with a Poker hand evaluator, and sent the results off to a GPT API to generate a summary of the situation and what action to take next. With the response, we used the RPI-native espeak library and RPI5 Bluetooth Library to connect to headphones and give the user their next play.

Challenges we ran into

Hardware is hard. We ran into a lot of cross-compatibility and dependency issues with all of the different RPI5 Python Libraries, and 90% of our time here at HoyaHacks was spent debugging, trying new solutions, having them fail, and looking for a new alternative. In the end, though, we figured things out.

Accomplishments that we're proud of

We're proud that we got so many different hardware and software components of our project to work together so well. We had to make a mid-hackathon trip to Microcenter, which was an achievement to get to in and of itself. We were all quite tired from the first week of a new semester, and we're proud of the work we did together to pull through.

What we learned

We went from zero to one with RPIs. From doing a basic login with SSH to manipulating files to sending requests to APIs to running autonomously, we learned A LOT about how these little computers work. And, how hard it is to make a functioning product with them. Very, very hard.

What's next for Ace

We'd like to hack some more peripherals onto the glasses to make the poker experience more interactive. Who knows, maybe at some point we'll end up hacking a pair of Metas.

Built With

• openai
• python
• raspberry-pi
• supabase