PolygonZone
https://github.com/roboflow/polygonzone- Category
- Developer Tools
- Rank
- No. 1708Tools index
Previous survey · No. 1716 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- roboflow
- GitHub
- 87 stars
- Date
About
A web utility to draw polygons and retrieve their coordinates for computer vision applications.
What it does
Drop an image into the browser, place vertices around the region you care about, then finish one or more shapes. PolygonZone turns those clicks into copyable NumPy and JSON coordinate data, with optional normalized values.
Why it's ranked here
PolygonZone solves one narrow, recurring setup task with little ceremony. Direct browser use, multiple polygons, editable points, normalized output, and clipboard-ready formats make it practical. Its deliberate scope also matters: it extracts geometry, but does not manage an annotation workflow.
What's good
The interaction supports polygon, line, and edit modes. Users can zoom, undo, discard unfinished work, clear shapes, and view live cursor coordinates. It accepts common web image formats, preserves coordinates relative to the original image scale, and offers both Python-oriented and JSON output.
Tradeoffs
The project explicitly targets desktop use and is not an annotation tool. It provides polygon coordinates, not labels, dataset management, review flows, or training integration. Image loading accepts PNG, JPEG, JPG, and WebP files. The supplied interface exposes browser interaction rather than a CLI, library, or API.
How to use it well
Use it when a computer vision script needs a few hand-chosen regions or lines from representative images. Draw, refine, copy the preferred coordinate format, then place that data in your application. Choose a dedicated annotation system when you need labeled examples, dataset organization, collaboration, or repeated production work.
Technical notes+
index.html defines a static canvas interface with polygon, line, and edit modes, clipboard controls, normalized-coordinate selection, fullscreen, undo, clearing, and image saving controls. script.js uses the Canvas 2D API, an off-screen canvas, FileReader, DOM events, and in-memory point arrays; it rescales pointer positions back to image coordinates, normalizes polygon winding, supports vertex dragging, and validates dropped image MIME types. styles.css supplies the desktop-oriented two-column layout and a limited responsive rule. CNAME maps the site to polygonzone.roboflow.com. LICENSE.md applies the MIT License.
Observed
- License
- MIT License
- Primary language
- JavaScript, with HTML and CSS
- Interface
- Static browser-based web application
- Install surface
- The README directs users to open the hosted application; it provides no installation command
- Platform support
- Designed for desktop use
- Export formats
- NumPy-style arrays and JSON
- Accepted image types
- PNG, JPEG, JPG, and WebP
- Deployment structure
- A custom domain is declared through a CNAME file
Read from README.md, script.js, CNAME, index.html, LICENSE.md, styles.css, CONTRIBUTING.md, .github/workflows/welcome.yml, .github/ISSUE_TEMPLATE/bug-report.yml, .github/ISSUE_TEMPLATE/feature-request.yml.
What it can do
Draw polygons on canvas
Mouse clicks and drag movements → Visual polygon shapes
Extract polygon coordinates
Drawn polygon shapes → Numerical coordinate data
Generate coordinate arrays for computer vision
Polygon boundaries → Formatted coordinate arrays
Create region of interest definitions
Polygon drawings → ROI coordinate specifications
Export polygon data
Completed polygon shapes → Exportable coordinate files
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.