MeMè | Devpost

💡 Inspiration

We were inspired by the childhood nostalgia and resurgence of collectibles!

Collecting has always been a part of childhood — from Pokémon cards to sports memorabilia — and it sparks imagination, strategy, and joy. But kids today don’t have a safe, playful place to manage their collections, understand their value, or learn about money in a fun, meaningful way. We wanted to give kids a digital playground for their collections, where every card they scan or trade becomes a lesson in budgeting, decision-making, and responsibility, while being guided by characters they love.

⚡️ What it does

MeMè turns collecting into an experience that teaches kids about money. Kids scan their cards or collectibles to add them to a digital collection, track their value, and organize wishlists. Parents can fund the wallet, letting kids manage spending safely. They can talk and listen to their favorite characters like Ash Ketchum, who offer tips on budgeting, making smart trades, and planning purchases. The app also provides AI-powered recommendations for new collectibles based on the kid's budget.

⚙️ How we built it

We designed the interface in Figma to be bright, playful, and kid-friendly.

Then we used the Flask API to integrate the backend with our frontend.

On the backend, we utilized a combination of tesseract text OCR and Gemini to accurately detect which card is being scanned along with any price tag on the card, if present.

We find the current value of a collection based on what the cards are reselling for online.

When a user selects an avatar, we use ElevenLabs to generate the voice for the character which gives AI-powered recommendations for what to purchase and additional budgeting tips.

We also integrated an open-source payment API (Stripe) to seamlessly load money.

🧠 Challenges we ran into

One of the hardest challenges we faced was integrating the OCR feature to scan properly with a single photo. Unlike a video stream which provides multiple image captures, the photo only provides one, which gives us significantly less samples of the card. Because of this, we had to use multiple grayscale, filtering, and image enhancement techniques to be able to consistently scan the collectable cards and identify them with high accuracy.

Integrating ElevenLabs for natural-sounding AI character voices also required tuning to feel playful without being overwhelming.

⭐️ Accomplishments that we're proud of

integrating the functionality to scan cards and add them to their collection, identifying if there is a price tag on it
kid-first interface that is intuitive, engaging, and immersive.
customized character voices for interactive learning

✍️ What we learned

Before this project, we had little experience with OCR and integrating tools like ElevenLabs and Stripe. By the end, we not only significantly improved our skills in these areas but also gained a better understanding of connecting multiple services and especially leveraging AI to enhance the capabilities of our application. Designing for two audiences, kids and parents, taught us the importance of perspective-switching and thoughtful UX, along with merging fun and education in a way that feels natural to kids.

🚀 What's next for MeMè

Moving forward, we plan to refine the financial literacy lessons that kids can learn from as well as creating a parental dashboard for additional safety measures. It would be amazing to support more genres of collectibles that are especially relevant to kids today, such as labubus and smiskis.