Pictonotes

Output 1
Output 2

Inspiration

Engineering students often struggle to grasp complex academic diagrams and handwritten structures quickly. We wanted to build a zero-friction tool that bridges the gap between static textbook diagrams and interactive learning, making revision fast and personalized.

What it does

Pictonotes allows users to drop any complex engineering diagram, formula sheet, or hand-drawn note, provide customizable steering text instructions (like "Explain like I'm 5" or "Summarize in 3 bullet points"), and instantly generates structured Markdown notes alongside an Interactive Sandbox for dynamic real-time workspace collaboration.

How we built it

Frontend Layer: Built a clean, lightweight, utility-first user interface using Streamlit.
Core AI Integration: Integrated Google Gemini API to handle multimodal image processing and context-driven text parsing.
Asynchronous Pipeline: Leveraged viaSocket webhooks to safely dispatch background logs and system payloads to external tracking dashboards in real time without hurting frontend latency.
Image Processing: Used Pillow (PIL) to stream and read device uploads safely.

Challenges we ran into Handling large multimodal payloads and formatting strict markdown structure dynamically was tricky, but we resolved it by engineering refined system prompts for Gemini and using native Streamlit container tools to split complex UI chunks.

What we learned We learned how to design a non-bloated, minimalist full-stack layout and realized the sheer power of asynchronous microservices like viaSocket webhooks when syncing data channels on the fly.