Question/Answer System for MLB

Application interface

Description - We have made an AI enabled solution to enhance baseball fans experience and increase their engagement with the sport. We have added features to allow it’s users to do a lot stuff with MLB videos and even get to know about the details of the players in that video

1.Interactive Q&A on Video Clips – Fans can ask questions about plays and get instant answers with video references, making the game more interactive.

Player Recognition – Instantly identify players with stats and historical data, strengthening fan connections.
Real-time Video Insights – AI-driven overlays provide live stats, predictions, and analytics, deepening fan engagement.
AI-Generated Highlights – Users can request key moments based on prompts, enabling quick and personalized game recaps.
Clip & Screenshot Downloads – Seamlessly capture and share game highlights, increasing fan-driven content and social interaction.

Open Source Git repo - https://github.com/GreymatterzOrg/Google_Hackathon_Challenge

Youtube Demo video - https://youtu.be/Ph8M-rqlCUU?si=cS1q_SM0p2CUmdVX

Inspiration

We have made an AI enabled solution to enhance baseball fans' experience and increase their engagement with the sport. We aspired to make a solution that would allow the valuable and passionate fans of the sport to personalize the way they want to connect with the sport. We have added quite a number of features for the fans to engage them and help them learn more about their favourite players.

What it does

Interactive Q&A on Video Clips – Fans can ask questions about plays and get instant answers, making the game more interactive and at the same time gain knowledge about the sport and the players.
Player Recognition – Instantly identify players with stats and historical data, strengthening fan connections to stay connected and learn about your favourite player.
Real-time Video Insights – Want real time insights about the uploaded video, our platform will provide real time insights on the video of your choice.
AI-Generated Best Moments – Seamlessly capture game clips and screenshots of the best moments of the uploaded according to the personalized prompt and the desired clip duration of the best moments video clips.
Clip & Screenshot Downloads – Effortlessly, save the best clips and screenshots according to your preferences in the gallery after creating a playlist .

How we built it

We used technologies like Python(flask) for backend integration and making the apis. Also through flask we have connected to the gemini model to use it for different prior mentioned functionalities and also through flask only we have made use of the api dataset provided by the MLB to get information about the player and other details.

For frontend development we have used React.js(Vite) which makes the frontend very lite and easily runnable on any platform and through React.js we have connected to the backend server and integrated all the functionalities.

For the real time communication we used socket to get real time insights without constantly calling the apis.

Challenges we ran into

We ran into a couple of challenges like how to use the provided dataset of the videos and the apis, but soon we figured out the appropriate apis and videos that aligned with the challenge we chose and the second problem we faced was how to enable real time communication between backend and frontend application but we figured it too and implemented it in the project.

Accomplishments that we're proud of

We were elated after the successful integration of such a vast dataset provided to us and implemented the complete idea for enhancing the baseball fans engagement and their personalized experience in such a short time duration.

What we learned

We learned more about both the technologies(Gemini and react) we used. Apart from that we gained knowledge about the data analysis and how to integrate such a vast dataset with our application. And most importantly we learned about such a beautiful sport.

What's next for Question/Answer System for MLB

From here we are planning to add more features to our platform including more statistics and deep dive into the dataset of apis and others provided and make our platform more informative yet user friendly through which all the fans of this amazing sport can stay connected with the sport and their favourite players.

Built With

api
dataset
gemini
mlb
python
react

Submitted to

Google Cloud x MLB(TM) Hackathon – Building with Gemini Models

Created by

I worked on the frontend to integrate apis and real time communication with the server through sockets for live insights. Also contributed to the design process and implementation.

sumit rout
I managed the project, gave feedback on UI/backend, went over iterations with team and testing with team, also provided feedback on video to the team.

Alok Dharayan
I played a key role in Python backend development and testing, focusing on seamless integration with MLB APIs and the Gemini model. My responsibilities included designing and implementing robust API endpoints to enhance the project's functionality, ensuring efficiency, scalability, and optimal performance. I also ensure the smooth real time communication to the frontend using the web sockets.

Aarif Saifi
I contributed to the Python backend development and testing. My responsibilities included integrating with MLB APIs and the Gemini model, as well as writing several API endpoints to support the project’s functionality.

Aarush Dharayan