Chickadee Studio：偵測網站隱藏及偽裝廣告的桌面工具開源
I built my first open-source project: Chickadee Studio — a desktop tool for detecting hidden and disguised advertising on websites
Chickadee Studio 是一款開源桌面工具，用於識別網站內容中可能隱藏或偽裝的廣告，涵蓋貼文、留言及圖片。它結合 PaddleOCR、DOM 與 CSS 分析、文字混淆偵測及本地 ML 模型，並提供截圖和證據供人工審查，亦支援可選的 AI 輔助分析。核心分析在本地執行，專案採用 Apache 2.0 授權。
Hey everyone!
I wanted to share my first open-source project,*Chickadee Studio.
It's a desktop application I built to help identify potentially hidden or disguised advertising in website content, including posts, comments, and images.
The idea came from a problem I found interesting: not all advertising is immediately visible or easy to identify with simple keyword matching.
Some content uses look-alike characters, unusual Unicode combinations, text embedded in images, or even hidden HTML elements.
So I wanted to experiment with combining several techniques to analyze this kind of content.
What it does:
- Scans webpages, posts, comments, and images
- Uses OCR (PaddleOCR) to extract text from images
- Analyzes DOM elements and CSS for potentially hidden content
- Detects text obfuscation techniques
- Uses local ML models to help identify advertising context
- Provides screenshots and evidence for manual review
- Supports optional AI-assisted analysis
Tech stack: Python, Electron, React, Playwright, PaddleOCR, and scikit-learn.
The core analysis runs locally, and the project is available under the Apache 2.0 license.
It's still an early project, and I'm sure there are plenty of things that could be improved.
Since this is my first open-source release, I'd especially appreciate feedback on the architecture, detection approach, usability, or anything else that stands out.
GitHub: https://github.com/chickadee-project/chickadee_studio
Thanks for checking it out!
來源：DEV Community · dev.to