We've conducted a study to measure the accuracy of the RealEye eye-tracking online tool. As a result, the following conclusions can be drawn:
- average accuracy is in the range of 90 to 156 px (depending on the place on the screen),
- average accuracy for all measurements is 113 px,
- accuracy range for participants (in total) varies from 60 px to over 200 px,
- the most accurate measurements are made for the middle part of the screen,
- the worst accuracy is obtained in the bottom corners of the screen,
- over half of the results have accuracy up to 100 px (and 75% up to 150 px),
- almost every result contains in the range of 0 - 300 px so that we can assume higher values as gross errors,
- gross errors aren’t common and don’t affect the accuracy,
- results were obtained from different participants (random people without knowledge about how the technology works or how important it is to follow the instructions) and indifferent environments like devices used, lightning conditions, head movements, etc., so they are realistic.
Read more here: RealEye Technical Whitepaper.
Please remember that the better hardware/software your participant has, the more accurate results you'll get. Check the recommended requirements.
In the following paper:
Katarzyna Wisiecka, Krzysztof Krejtz, Izabela Krejtz, Damian Sromek, Adam Cellary, Beata Lewandowska, and Andrew Duchowski. 2022. Comparison of Webcam and Remote Eye Tracking. In 2022 Symposium on Eye Tracking Research and Applications (ETRA' 22). Association for Computing Machinery, New York, NY, USA, Article 32, 1–7. https://doi.org/10.1145/3517031.3529615
we compared measurement error and validity of the RealEye webcam solution to remote (the GP3) eye-tracking and an integrated method in point detection tasks. We measured the error defined as a distance (in pixels) between the center of the AOI and the position of the participant's fixation. In laboratory conditions, this distance was:
- webcam - 45.1,
- remote - 34.2,
- integrated condition - 27.70.