Abstract
Eye-tracking calibration models are often specific to a particular device’s sensor geometry, requiring extensive data collection and retraining when camera or LED positions are altered in new hardware. This process is resource-intensive and hinders the portability of models across different devices. A method is disclosed for adapting a comprehensive eye-tracking model from a reference device to a new device with a different sensor layout. By collecting a minimal set of new calibration points on the target device, a transformation matrix is generated. This matrix remaps the existing reference model to the new geometry. The primary purpose of this technique is to enable rapid deployment of eye-tracking on new hardware configurations by reusing existing models, thereby bypassing the need for a complete data-collection and training process.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Franco, Mar Gonzalez; Gonzalez, Eric Jordan; Abramyan, Lucy; Ahuja, Karan; Gurumurthy, Prasanthi; and Patel, Khushman Jayantilal, "Cross-Device Eye-Tracking Calibration via Reference Model Remapping", Technical Disclosure Commons, (December 15, 2025)
https://www.tdcommons.org/dpubs_series/9032