About

Prototype

ABOUT

DRISHTI

DRISHTI is a document image-quality screening prototype. It looks at a photo or scan of a document and reports nine quality signals so a person can retake a bad image before submitting it.

This build demonstrates the interface and the checking pipeline. It is not an identity-verification service.

Prototype disclaimer. Quality thresholds are experimental, results are advisory, and nothing here confirms who a document belongs to. Do not upload real identity documents.

Prototype facts

Sample values for this demonstration build.

Version
0.4.0 (prototype)
Quality checks
Nine
Accepted formats
JPEG, PNG, WebP
Maximum upload
10 MB
Analysis modes
Photo and scan
Interface
UX4G design system
Accounts
None
Image storage
None

How it is built

INTERFACE

UX4G web components

Plain HTML, CSS and JavaScript styled with the UX4G design system. Light and dark themes come from the same token set.

SERVICE

Python analysis API

A small FastAPI service validates the upload and runs the nine quality modules. It is public and has no rate limiting.

Data and privacy

  • Images are analysed in memory and are not written to disk.
  • Theme and text size are the only values stored in your browser, under a single preferences key.
  • There is no account, sign-in or password storage in this build.
  • The screening history you see is sample data, not a real record store.

Known limitations

  • Camera capture is a placeholder; upload an existing image.
  • Screening history is not persisted between visits.
  • The analysis endpoint is unauthenticated and unrated, so it should only be used with synthetic images.
  • Thresholds have not been calibrated against a real document set.

Where to next