Project overview
Building Circo
A Windows desktop project that turns a selection into an image search, editable text, or a translation.
ElectronWindows OCRFirebase

The problem
Useful information gets stuck in pixels
Text can be embedded in a screenshot or video. An object may be easy to recognize and hard to describe. Circo gives both cases the same starting point: select the part of the screen you care about, then choose what to do with it.
The capture overlay keeps the selected region in context while the user chooses an action. A floating control and configurable keyboard shortcut keep that workflow available across applications.
Implementation
From screen coordinates to a useful result
- Capture and display scaling
- The Electron main process captures the active display. The overlay maps logical selection coordinates to the captured image’s physical pixels, including on mixed-DPI setups.
- Local recognition
- A C# helper invokes Windows OCR. The app converts recognized word positions into selectable text at the original location.
- Online actions and account state
- Visual search forwards only the selected image to Google Lens. Translation sends extracted text to Azure Translator. Firebase handles sign-in and account entitlements.
- Recovery and verification
- Regression tests cover session renewal, failed history writes, and payment confirmation retries. The website uses static HTML and loads product videos when requested.
Contact
Discuss the project
Get in touch about Circo’s interface, desktop implementation, or technical decisions.
Email Circo