get the analysis of the image
select the object you want to know about
remix the object with your desired choice

BananaLens — See Beyond the Image 🍌

Inspiration

Ig images more interactive by building an AI-powered visual assistant that helps people discover, undmages contain valuable information, but understanding everything in a picture can take time. We wanted to make explorinerstand, and transform what they see.

What It Does

BananaLens turns an ordinary image into an interactive AI experience. Users can:

Detect Objects: Upload a JPG or PNG image and identify objects with visual bounding boxes.
Understand Objects: Explore detected objects and get AI-generated descriptions.
Ask Questions: Ask BananaLens questions about an uploaded image and receive contextual answers.
Remix Visuals: Use AI-powered image remixing to creatively transform visual ideas.
Explore Interactively: Select objects, revert changes, switch languages, and download results.

How We Built It

BananaLens combines a React and Vite frontend with a Python FastAPI backend. Google Gemini powers image understanding and question answering, while Pollinations provides the image-remixing capability. The application connects these services through API endpoints and is deployed on Vercel.

Challenges We Faced

One of our main challenges was connecting the frontend and backend in a production environment. We worked through deployment configuration, environment variables, API connectivity, and cross-origin resource sharing (CORS). Integrating multiple AI services while maintaining a smooth user experience was another important challenge.

What We Learned

Building BananaLens strengthened our understanding of full-stack development, REST APIs, AI integration, image processing, cloud deployment, and debugging production issues. We also learned how important it is to make AI capabilities accessible through a simple, interactive interface.

What's Next

We aim to improve detection accuracy, expand language support, refine image-editing controls, and introduce more powerful visual understanding features.

BananaLens makes images more than something you look at — it makes them something you can explore.We're proud to have built BananaLens into a working, deployed AI-powered visual assistant that combines image understanding, object detection, conversational AI, and creative image remixing in one application.

Some of our key accomplishments include:

AI-powered object detection: Identifying objects in images and displaying visual bounding boxes.
Interactive image understanding: Enabling users to select objects and ask questions about their images.
Creative AI integration: Connecting Google Gemini for visual understanding and Pollinations for image remixing.
Full-stack development: Building a React frontend and FastAPI backend that work together as a complete application.
Successful cloud deployment: Deploying both the frontend and backend on Vercel so users can access BananaLens online.
Overcoming real-world challenges: Troubleshooting API connectivity, environment variables, CORS, and deployment issues to bring the project online.

Most importantly, we're proud that BananaLens brings understanding and creativity together in a single, interactive experience. This project pushed us to learn, experiment, debug, and turn an idea into a working application under hackathon time constraints.