CogniVerse

• 

  Student View: An interactive 3D classroom. Students are notified of new uploads, can access materials directly from virtual bookshelves

• 

  Messy handwritten notes automatically transformed into formatted LaTeX PDFs, making advanced contents (algorithms, math proofs) professional

• 

  Student View: A searchable library of all uploaded study materials, allowing students to easily browse, filter, and access knowledge

• 

  Educator View: Educators upload handwritten notes, convert them into organized study materials,easily manage all course content in one place

• 

• 

• 

• 

Inspiration

Educators and students alike spend far too much time on the mechanics of formalizing and distributing learning materials. Notes get stuck in messy handwriting, ideas remain half-finished, and valuable time is lost converting them into polished documents. LaTeX produces beautiful output, but it demands expertise and hours of formatting. Word processors are simple, but lack the elegance and precision needed for mathematics, algorithms, and even structured writing in the humanities. We wanted a system where anyone with an idea could capture it quickly and have it automatically transformed into professional, shareable material inside an interactive learning space.

What it does

Our platform converts even the messiest handwriting into neatly formatted LaTeX code and PDF output, handling mathematical symbols, algorithms, and prose with equal fluency. Whether you are writing a book chapter, drafting lecture notes, or sketching a study guide, the tool automates the process of producing publication-quality documents. Students and educators then experience these materials in an interactive 3D classroom built with \textbf{three.js}, where textbooks live on virtual bookshelves. This transforms learning from static PDFs into an immersive, collaborative environment, making knowledge more accessible, engaging, and alive.

How we built it

We combined OCR and LLM-based models to translate handwritten input into structured LaTeX code, which is then compiled into clean, professional PDFs. The 3D classroom was built using \textbf{three.js}, providing an immersive interface for browsing and interacting with documents.

Challenges we ran into

• Our initial OCR tool, Tesseract, struggled with messy handwriting and especially with mathematical notation.
  
• Ensuring consistent and valid LaTeX output from Cohere’s LLM required custom prompting and validation.
  
• Generating PDFs from LaTeX presented multiple trade-offs:
  
  • latexonline.cc: convenient but required hosting the input files on a server.
    
  • KaTeX: fast web rendering but syntax differs from LaTeX.
    
  • pdflatex: reliable local compilation but requires full LaTeX installation.

Accomplishments that we're proud of

• Built an interactive 3D classroom interface in three.js.
  
• Automated end-to-end workflow: from raw handwriting to polished PDF.
  
• Created a cross-disciplinary tool useful for both STEM and humanities.
  

What we learned

We discovered the complexity of bridging handwritten recognition with symbolic formalisms like LaTeX, and the importance of designing interfaces that educators actually want to use. We also learned that immersive visualization adds value, not just as a novelty, but as a way to reimagine how learning materials are shared and experienced.

What's next for Cogniverse

• Multi-language support for global classrooms.
  
• Richer interactivity: annotation, collaboration, and AI-driven tutoring inside the 3D environment.
  
• Expansion beyond education-drafting books, research articles, or creative projects

Built With

• cohere
• fastapi
• javascript
• python
• three.js