AP LogoAustin Profenius
Web App

ICEBURG

Award-winning Chrome extension built at Hack@Davidson that integrates YouTube transcript extraction, page content analysis, a click-and-drag screenshot tool with Tesseract OCR, and Flask WebSockets for real-time streaming — all powered by Cloudflare.

Overview

ICEBURG is a Chrome extension built in 48 hours at the Hack@Davidson hackathon, where it won Best AI Project for its Cloudflare-powered content analysis capabilities.

The extension extracts YouTube transcripts via the YouTube API, pulls page content (tested with 100K+ characters) for analysis, and includes a custom JavaScript click-and-drag screenshot tool that feeds into Tesseract OCR for dynamic text extraction from images.

A Flask backend with WebSockets streams real-time results and UI updates back to the extension as content is being processed.

error

The Problem

  • Extracting useful information from web pages, YouTube videos, and images requires switching between multiple separate tools
  • Long-form page content and video transcripts need to be analyzed quickly without manual copy-pasting
  • OCR tools typically require uploading files to external services rather than capturing content directly in the browser
check_circle

The Solution

  • Built a unified Chrome extension that handles YouTube transcripts, page content extraction, and screenshot OCR in one tool
  • Integrated the YouTube API to automatically extract and analyze video transcripts
  • Engineered a JavaScript click-and-drag screenshot tool that captures arbitrary screen regions and passes them to Tesseract OCR
  • Implemented Flask WebSockets to stream processing results and UI updates back to the extension in real time

Technical Implementation

YouTube Transcript Extraction

The extension hooks into YouTube pages to extract full video transcripts via the YouTube API, making them available for analysis and summarization without leaving the browser.

Click-and-Drag Screenshot + OCR

A custom JavaScript tool lets users click and drag to select any region of the page. The captured screenshot is sent to Tesseract OCR on the Flask backend for text extraction.

Flask WebSocket Streaming

The Flask backend uses WebSockets to stream real-time processing results and UI updates back to the Chrome extension, providing immediate feedback as content is analyzed.

Large Content Handling

The page content extractor was tested with 100K+ characters of content, ensuring reliable extraction and analysis of long-form articles, documentation, and other text-heavy pages.