This study presents the first data-driven review of research on the privacy of eye tracking, covering gaze, iris, and eye image data across immersive, mobile, and clinical contexts. The analysis examined 78 papers published between 2015 and 2025 using ensemble topic modeling with non-negative matrix factorization. Nine topics emerged, showing how normative and technical approaches address privacy across different stages of eye tracking data processing. Findings reveal that normative works emphasize inference of identity and personal traits as an unresolved risk, whereas technical studies treat privacy as a computational problem and mitigate risks within specific contexts. However, protections are often constrained by utility requirements, meaning eye tracking data cannot be sufficiently safeguarded without impairing function, leaving enough detail to enable inference. This persistent vulnerability suggests that eye tracking data may warrant regulatory protection comparable to other sensitive categories, requiring stronger governance and safeguards across its collection and use.

This study investigates gaze behavior as a potential biomarker for individuals with Social Anxiety Disorder during immersive virtual reality socio-evaluative tasks. Thirty participants diagnosed with SAD and thirty matched healthy controls completed three VR tasks while eye movements were recorded. Gaze features were analyzed using a neural predictive model with PAC-Bayes estimating generalization and uncertainty. The model predicted group membership with accuracies of 87 % and 93 % for performance and social tasks, with certified PAC-Bayes lower bounds accuracy of 78 % and 84 %, respectively. Variations in social demand across tasks were associated with corresponding differences in predictive gaze features. Gender bias was investigated, but was not detected. Model predictions correlated strongly with standardized psychometric measures. These findings indicate that VR-based eye tracking, combined with machine learning models, provides a scalable, explainable, and objective method to assess SAD.

This paper presents PyEtSimul, an open-source Python-based framework for simulating video-based eye trackers by generating synthetic eye features through geometric modeling. The framework allows flexible positioning of eyes, cameras, and light sources in 3D space, with controlled variation of eye anatomical features and camera properties. PyEtSimul generalizes corneal modeling by representing the cornea as a conic surface rather than the common sphere. It also supports non-circular pupil shapes, size-dependent pupil decentration, eyelid occlusion, and camera lens distortion. It supports systematic data generation and principled comparison of gaze estimation algorithms across calibrated and uncalibrated settings. These features enable analyses not possible with other available simulators. PyEtSimul facilitates controlled experiments with known parameters often latent in normal settings, enabling reproducible benchmarking and systematic exploration of hardware designs. By generating fully synthetic data, PyEtSimul removes privacy concerns and the need for costly hardware, making it practical for both educational and research applications.

We propose to use unlabelled eye image data for domain adaptation of an iris segmentation network. Adaptation allows the model to be less reliant on its initial generality. This is beneficial due to the large variance exhibited by eye image data which makes training of robust models difficult. The method uses a label prior in conjunction with network predictions to produce pseudo-labels. These are used in place of ground-truth data to adapt a base model. A fully connected neural network performs the pixel-wise iris segmentation. The base model is trained on synthetic data and adapted to several existing datasets with real-world eye images. The adapted models improve the average pupil centre detection rates by 24% at a distance of 25 pixels. We argue that the proposed method, and domain adaptation in general, is an interesting direction for increasing robustness of eye feature detectors.

The duration of the so-called "Quiet Eye" (QE) - the final fixation before the initiation of a critical movement - seems to be linked to better perceptual-motor performances in various domains. For instance, experts show longer QE durations when compared to their less skilled counterparts. The aim of this paper was to replicate and extend previous work on the QE [Vickers and Williams 2007] in elite biathletes in an ecologically valid environment. Specifically, we tested whether longer QE durations result in higher shooting accuracy. To this end, we developed a gun-mounted eye tracker as a means to obtain reliable gaze data without interfering with the athletes' performance routines. During regular training protocols we collected gaze and performance data of 9 members (age 19.8 ± 0.45) of the German national junior team. The results did not show a significant effect of QE duration on shooting performance. Based on our findings, we critically discuss various conceptual as well as methodological issues with the QE literature that need to be aligned in future research to resolve current inconsistencies.

We propose a binocular model-based method that only uses a single camera and an infrared light source. Most gaze estimation approaches are based on single eye models and with binocular models they are addressed by averaging the results from each eye. In this work, we propose a geometric model of both eyes for gaze estimation. The proposed model is implemented and evaluated in a simulated environment and is compared to a binocular model-based method and polynomial regression-based method with one camera and two infrared lights that average the results from both eyes. The method performs on par with methods using multiple light sources while maintaining robustness to head movements. The study shows that when using both eyes in gaze estimation models it is possible to reduce the hardware requirements while maintaining robustness.

Preprocessor directives (#ifdefs) are often used to implement compile-time variability, despite the critique that they increase complexity, hamper maintainability, and impair code comprehensibility. Previous studies have shown that the time of bug finding increases linearly with variability. However, little is known about the cognitive process of debugging programs with variability. We carry out an experiment to understand how developers debug programs with variability. We ask developers to debug programs with and without variability, while recording their eye movements using an eye tracker. The results indicate that debugging time increases for code fragments containing variability. Interestingly, debugging time also seems to increase for code fragments without variability in the proximity of fragments that do contain variability. The presence of variability correlates with increase in the number of gaze transitions between definitions and usages for fields and methods. Variability also appears to prolong the "initial scan" of the entire program that most developers initiate debugging with.

We investigate the gaze estimation error induced by pupil size changes using simulated data. We investigate the influence of pupil diameter changes on estimated gaze point error obtained by two gaze estimation models. Simulation data show that at wider viewing angles and at small eye-camera distances, error increases with increasing pupil sizes. The maximum error recorded for refracted pupil images is 2.4° of visual angle and 1.5° for non-refracted pupil projections.

We propose a implicit calibration method for estimating the offset between the optical and visual axes without active participation of a user. The method relies on a fully calibrated setup and uses the relation between the optical axes of both eyes, the position of the center of the cornea, and the position of the display. The method is based on the assumption that the visual axes of both the eyes coincide on the display. The implicit calibration method estimates the offsets by continuously estimating the angle kappa through kernel density estimation. From the numerical simulation, the accuracy of our method is comparable or better than existing methods for implicit calibration.

A novel normalization principle for robust glint detection is presented. The method is based on geometric properties of corneal reflections and allows for simple and effective detection of glints even in the presence of several spurious and identically appearing reflections. The method is tested on both simulated and data obtained from web cameras. The proposed method is a possible direction towards making eye trackers more robust to challenging scenarios.

This paper presents a fully calibrated extended geometric approach for gaze estimation in three dimensions (3D). The methodology is based on a geometric approach utilising a fully calibrated binocular setup constructed as a head-mounted system. The approach is based on utilisation of two ordinary web-cameras for each eye and 6D magnetic sensors allowing free head movements in 3D. Evaluation of initial experiments indicate comparable results to current state-of-the-art on estimating gaze in 3D. Initial results show an RMS error of 39-50 mm in the depth dimension and even smaller in the horizontal and vertical dimensions regarding fixations. However, even though the workspace is limited, the fact that the system is designed as a head-mounted device, the workspace volume is relatively positioned to the pose of the device. Hence gaze can be estimated in 3D with relatively free head-movements with external reference to a world coordinate system and is therefore offering flexibility and movability within certain constraints.

A novel method for video-based head gesture recognition using eye information by an eye tracker has been proposed. The method uses a combination of gaze and eye movement to infer head gestures. Compared to other gesture-based methods a major advantage of the method is that the user keeps the gaze on the interaction object while interacting. This method has been implemented on a head-mounted eye tracker for detecting a set of predefined head gestures. The accuracy of the gesture classifier is evaluated and verified for gaze-based interaction in applications intended for both large public displays and small mobile phone screens. The user study shows that the method detects a set of defined gestures reliably.

This paper investigates the parallax error, which is a common problem of many video-based monocular mobile gaze trackers. The parallax error is defined and described using the epipolar geometry in a stereo camera setup. The main parameters that change the error are introduced and it is shown how each parameter affects the error. The optimum distribution of the error (magnitude and direction) in the field of view varies for different applications. However, the results can be used for finding the optimum parameters that are needed for designing a head-mounted gaze tracker. It has been shown that the difference between the visual and optical axes does not have a significant effect on the parallax error, and the epipolar geometry can be used for describing the parallax error in the HMGT.

This paper presents a low-cost gaze tracking system that is based on a webcam mounted close to the user's eye. The performance of the gaze tracker was evaluated in an eye-typing task using two different typing applications. Participants could type between 3.56 and 6.78 words per minute, depending on the typing system used. A pilot study to assess the usability of the system was also carried out in the home of a user with severe motor impairments. The user successfully typed on a wall-projected interface using his eye movements.

Homography normalization is presented as a novel gaze estimation method for uncalibrated setups. The method applies when head movements are present but without any requirements to camera calibration or geometric calibration. The method is geometrically and empirically demonstrated to be robust to head pose changes and despite being less constrained than cross-ratio methods, it consistently performs favorably by several degrees on both simulated data and data from physical setups. The physical setups include the use of off-the-shelf web cameras with infrared light (night vision) and standard cameras with and without infrared light. The benefits of homography normalization and uncalibrated setups in general are also demonstrated through obtaining gaze estimates (in the visible spectrum) using only the screen reflections on the cornea.

Despite active research and significant progress in the last 30 years, eye detection and tracking remains challenging due to the individuality of eyes, occlusion, variability in scale, location, and light conditions. Data on eye location and details of eye movements have numerous applications and are essential in face detection, biometric identification, and particular human-computer interaction tasks. This paper reviews current progress and state of the art in video-based eye detection and tracking in order to identify promising techniques as well as issues to be further addressed. We present a detailed review of recent eye models and techniques for eye detection and tracking. We also survey methods for gaze estimation and compare them based on their geometric properties and reported accuracies. This review shows that, despite their apparent simplicity, the development of a general eye detection technique involves addressing many challenges, requires further theoretical developments, and is consequently of interest to many other domains problems in computer vision and beyond.

This paper presents StarGazer - a new 3D interface for gaze-based interaction and target selection using continuous pan and zoom. Through StarGazer we address the issues of interacting with graph structured data and applications (i.e. gaze typing systems) using low resolution eye trackers or small-size displays. We show that it is possible to make robust selection even with a large number of selectable items on the screen and noisy gaze trackers. A test with 48 subjects demonstrated that users who have never tried gaze interaction before could rapidly adapt to the navigation principles of StarGazer. We tested three different display sizes (down to PDA-sized displays) and found that large screens are faster to navigate than small displays and that the error rate is higher for the smallest display. Half of the subjects were exposed to severe noise deliberately added on the cursor positions. We found that this had a negative impact on efficiency. However, the user remained in control and the noise did not seem to effect the error rate. Additionally, three subjects tested the effects of temporally adding noise to simulate latency in the gaze tracker. Even with a significant latency (about 200 ms) the subjects were able to type at acceptable rates. In a second test, seven subjects were allowed to adjust the zooming speed themselves. They achieved typing rates of more than eight words per minute without using language modeling. We conclude that the StarGazer application is an intuitive 3D interface for gaze navigation, allowing more selectable objects to be displayed on the screen than the accuracy of the gaze trackers would otherwise permit.

Recent developments in image search has made them sufficiently efficient to be used in real-time applications. GPS has become a popular navigation tool. While GPS information provide reasonably good accuracy, they are not always present in all hand held devices nor are they accurate in all situations, for example in urban environments. We propose a system to provide location-based services using image searches without requiring GPS. The goal of this system is to assist tourists in cities with additional information using their mobile phones and built-in cameras. Based upon the result of the image search engine and database image location knowledge, the location is determined of the query image and associated data can be presented to the user.

Low cost eye tracking has received an increased attention due to the rapid developments in tracking hardware (video boards, digital camera and CPU's) [Hansen and Pece 2005; OpenEyes 2005]. We present a gaze typing system based on components that can be bought in most consumer hardware stores around the world. These components are for example cameras and graphics cards that are made in large quantities. This kind of hardware differs from what is often claimed to be "off-the-shelf components", but which in fact is hardware only available from particular vendors.Institutions that supply citizens with communication aids may be reluctant to invest large amounts of money in new equipment that they are unfamiliar with. Recent investiagtions estimate that less than 2000 systems have actually been used by Europeans, even though more than half a million disabled people in Europe could potentially benefit from it. The main group of present users consists of people with motor neuron disease (MND) and amyotrophic lateral sclerosis (ALS). If the price of gaze communication systems can be lowered, it could become a preferred means of control for a large group of people [Jordansen et al. 2005]. Present commercial gaze trackers e.g. [Tobii 2005; LC-Technologies 2004] are easy to use, robust and sufficiently accurate for many screen-based applications but their costs exceed the budget of most people.We use a standard uncalibrated 400$ Sony consumer camera (Sony handycam DCR-HC14E) to obtain the image data. The camera is stationary and placed on a tripod close (variable) to the monitor, but the geometry of the user, monitor and camera varies among sequences. However, the users are sitting about 50 - 60 cm away from a 17" screen. A typical example of the setup is shown in figure 1. We use Sony standard video option for 'night vision' to create an glint with the build-in IR light emitter.Eye tracking based on common components is subject to several unknown factors as various system parameters (i.e. camera parameters and geometry) are unknown. Algorithms that employ robust statistical principles to accommodate uncertainties in image data as well as in gaze estimates in the typing process are therefore needed. We propose to use the RANSAC algorithm [Fischler and Bolles 1981] for both robust maximum likelihood estimation of iris observations [Hansen and Pece 2005] as well as for handling outliers in the calibration procedure [Morimoto et al. 2000].Our low-resolution gaze tracker can be calibrated in less than 3 minutes by looking at 9 predefined positions on the screen. The users sit on a standard office chair without headrests or other physical constraints. Under these conditions we have succeeded in tracking the gaze of people, obtaining accuracies about 160 pixels on screen. This is still less than accuracies claimed by the best current off-the-shelf eye trackers systems (i.e. 30-60 pixels). However comparing these eye trackers wouldn't be correct as they are based on different hardware and image data.Low-cost gaze trackers do not need to be as accurate and robust as the commercial systems, if they are used together with applications designed to tolerate noisy inputs.We use the GazeTalk [COGAIN 2005] typing communication system components and have through proper design of the typing interface, reduced the need for high accuracy. We have observed typing speeds in the range of 3 - 5 words per minute for untrained subjects using large on-screen buttons and a new noise tolerant dwell-time principle. We modify the traditional dwell-time activation to one that maintains a full distribution of all hypothetical button selections and then activate one button when the evidence become high enough.

In this paper we propose a log likelihood-ratio function of foreground and background models used in a particle filter to track the eye region in dark-bright pupil image sequences. This model fuses information from both dark and bright pupil images and their difference image into one model. The tracker overcomes the issues of prior selection of static thresholds during the detection of feature observations in the bright-dark difference images. The auto-initialization process is performed using cascaded classifier trained using adaboost and adapted to IR eye images. Experiments show good performance in challenging sequences with test subjects showing large head movements and under significant light changes.

What if eye trackers could be downloaded and used immediately with standard cameras connected to a computer, without the need for an expert to setup the system? This has already the case for head trackers, so why not for eye trackers?Using components off-the-shelf (COTS) for camera-based eye tracking tasks has many advantages, but it certainly introduces several new problems as less assumptions on the system can be made. As a consequence of using COTS the price for eye tracking devices can be reduced while increasing the accessibility of these systems. Eye tracking based on COTS holds potential for a large number of possible applications such as in the games industry and eye typing [Majaranta and Räihä 2002]. Different cameras may be purchased depending on the need and the amount of money the user is willing to spend on the camera. In this framework it is not possible to use IR light sources and other novel engineered devices as they cannot be bought in a common hardware store. Very little control over the cameras and the geometry of the setup can be expected. The methods employed for eye tracking should therefore be able to handle changes in light conditions and image defocusing and scale changes [Hansen and Pece 2003]. On the same token pan-and-tilt cameras cannot be used, thus forcing such systems to be passive. Figure 1 shows a possible setup of a COTS-based eye tracker. When designing systems for the general public, it is unrealistic to assume that people are able to do camera calibration and make accurate setups of camera, monitor and user. Since little is known about the setup, would this then require a vast amount of calibration points needed for gaze estimation? That is, how many calibration points are really needed? Obviously the more calibration points are used the better the chances are to be able to infer the mapping from the image to gaze direction. It would even be possible to sample the entire function space provided sufficiently many calibration points are given. From the point of view of the users, a low number of calibration points is preferred as calibration may be considered as a tedious procedure. Systems that require many calibration points for every session are therefore not likely to succeed. It is also important to know the accuracy in gaze determination when using COTS to determine their applicability for various tasks.

We propose a non-intrusive eye tracking system intended for the use of everyday gaze typing using web cameras. We argue that high precision in gaze tracking is not needed for on-screen typing due to natural language redundancy. This facilitates the use of low-cost video components for advanced multi-modal interactions based on video tracking systems. Robust methods are needed to track the eyes using web cameras due to the poor image quality. A realtime tracking scheme using a mean-shift color tracker and an Active Appearance Model of the eye is proposed. It is possible from this model to infer the state of the eye such as eye corners and the pupil location under scale and rotational changes.