[✨ FEATURE]: Option to bypass eye-tracker calibration using a valid calibration payload (for dev/admin testing)

[sohampirale]

Feature Description 📝

• Currently, when testing unmoderated tests with the eye tracker, calibration is mandatory every time a test is started.

• This feature proposes adding an option (or script-based mechanism) to bypass the calibration step by injecting a valid gaze calibration & evaluation result payload.

• This would allow contributors or admins to directly test the eye-analysis flow without repeatedly going through the calibration process.

Motivation 🌟

• While calibration is essential for real user sessions, during development and testing it becomes repetitive—especially when validating UI flows, reports, or backend processing.

• Having a way to reuse or inject a valid calibration payload would significantly speed up development and debugging.
  This could also be useful for admins who may need to test eye-tracking workflows frequently.

Expected Behavior 🤔

• Provide a way (flag, script, or admin-only option) to bypass calibration
• Allow injecting a valid gaze calibration & evaluation result payload
• The system should proceed as if calibration was completed successfully
• This option should be clearly marked as for development/admin use only
• Result quality may vary, but should be sufficient for testing and debugging purposes
• Script in eye-tracker-api or integration into web-eye-tracker-front

Additional Information ℹ️

This feature is not intended to replace calibration for real user sessions
It is mainly aimed at contributors, maintainers, and admins
Could be implemented as:

• A dev-mode toggle
• An admin-only endpoint
• A reusable calibration payload/script for testing

Also added short script and tested it on linear regression:

1.We need to give sessionId everytime

Screencast.from.2026-02-02.23-03-35.webm

or/and

2.Caching the caliberation results for per admin/dev either in DB or localstorage and then using them in web-eye-tracker-front (optionally)

Happy to help explore implementation details or constraints if this sounds useful 🙂

[sohampirale]

I am working on this