MARSH

Inspiration

In daily life, every user deals with perplexing digital displays like permissions, alerts, or errors. We wanted to make technology easy to grasp—for everybody, not just experts

What it does

To break down confusing screenshots, MARSH employs simple language and an AI-backed logic system. Users can provide an image path, and instant results will be received for clarity and safety levels.

How we built it

We developed MARSH with AI-generated code in Python and OCR classification rules. Gemini was employed for generating and optimizing the basic logic and explanation flow.

Challenges we ran into

Managing different formats of screenshots and obtaining clean text was difficult. We solved this by improving handling of OCR, which included adding fallback rules.

Accomplishments that we're proud of

developed a fully working prototype within the time limit of a hackathon. Solution runs consistently in terminal and clearly demonstrates real-world impact.

What we learned

We learned how AI-generated code can speed up development many times over. We also learned about the importance of clarity and the safety of users in digital experiences.