Gaze prediction is essential for addressing motion-to-photon latency and ensuring seamless foveated rendering in Virtual Reality. The reliability of gaze forecasting is highly sensitive to individual differences and the eye movements being predicted. We evaluate recurrent, transformer-based, and classification-guided architectures to assess their generalization capabilities across oculomotor events. Using the GazeBase VR and Meta Quest Pro datasets, we analyzed the relationship between the median (P50) and high-percentile (P95) error profiles across subjects. The analysis reveals significant performance variability, showing that subjects with low P50 errors do not always exhibit the lowest extreme-case errors. Consequently, low median errors do not guarantee the robustness of the utilized solution. We discuss inference performance and address the class imbalance problem in short-term gaze prediction. These results identify a gap in standardized evaluation methods, necessitating a shift toward P95-focused, subject-specific metrics to develop reliable and perceptually stable gaze-contingent systems.

We present a privacy-enhancing mechanism for gaze signals using a latent-noise autoencoder that prevents users from being re-identified across play sessions without their consent, while retaining the usability of the data for benign tasks. We evaluate privacy-utility trade-offs across biometric identification and gaze prediction tasks, showing that our approach significantly reduces biometric identifiability with minimal utility degradation. Unlike prior methods in this direction, our framework retains physiologically plausible gaze patterns suitable for downstream use, which produces favorable privacy-utility trade-off. This work advances privacy in gaze-based systems by providing a usable and effective mechanism for protecting sensitive gaze data.

This paper investigates the feasibility of fusing two eye-centric authentication modalities—eye movements and periocular images—within a calibration-free authentication system. While each modality has independently shown promise for user authentication, their combination within a unified gaze-estimation pipeline has not been thoroughly explored at scale. In this report, we propose a multimodal authentication system and evaluate it using a large-scale in-house dataset comprising 9202 subjects with an eye tracking (ET) signal quality equivalent to a consumer-facing virtual reality (VR) device. Our results show that the multimodal approach consistently outperforms both unimodal systems across all scenarios, surpassing the FIDO benchmark. The integration of a state-of-the-art machine learning architecture contributed significantly to the overall authentication performance at scale, driven by the model’s ability to capture authentication representations and the complementary discriminative characteristics of the fused modalities.

Eye movement biometric has recently reached a meaningful performance threshold within a gaze estimation pipeline. Prior research claimed that good biometric performance can be achieved from a relatively large set of weakly intercorrelated features with high temporal persistence (indexed by the measurement of reliability). In this study, we revisit this hypothesis in the context of a modern deep learning (DL)-based eye movement biometric system, using a publicly available eye-movement dataset. Specifically, we investigate whether the measurement of reliability of learned embeddings continues to predict biometric performance in DL-based biometrics. Our results confirm that temporal persistence—quantified by measurement of reliability—is a significant predictor of performance in DL-based biometric systems, extending prior findings into the DL-based biometric. We also examine how manipulating eye-tracking signal quality descriptors impacts the temporal persistence of embeddings, finding that degradation of any kind undermines their temporal persistence. As a general matter, we found that measurement of reliability is an important predictor of DL-based biometric performance, and also that DL-learned embeddings are generally weakly intercorrelated.

Virtual reality (VR) sensors capture large amounts of user data, including body motion and eye tracking, that contain personally identifying information. While privacy-enhancing techniques can obfuscate this data, incomplete privacy protections risk privacy leakage, which may allow adversaries to leverage unprotected data to identify users without consent. This work examines the extent to which unprotected body motion data can undermine privacy protections for eye tracking data, and vice versa, to enable user identification in VR. These findings highlight a privacy consideration at the intersection of eye tracking and VR, and emphasize the need for privacy protections that address these technologies comprehensively.

We present an analysis of the eye tracking capabilities of the Meta Quest Pro virtual reality headset using a dataset of eye movement recordings collected from 78 participants. We highlight the potential differences in user experience as a function of device performance using a novel, user-centered evaluation framework for eye tracking signal quality analysis. In addition to presenting classical signal quality metrics such as spatial accuracy and spatial precision, we also explore how spatial accuracy varies across the field of view for different users across the performance range of the device. This work contributes to a growing understanding of eye tracking signal quality in virtual reality platforms, where the usability of gaze-based applications is directly dependent on the quality of the device’s eye tracking signal.

Gaze-based user interfaces and interactions are becoming more prevalent in augmented and virtual reality (AR/VR). The effectiveness of eye tracking for interaction depends on its quality. Many studies discuss eye tracking as an input and interaction modality but do not provide details about eye tracking quality, making it difficult to compare findings. Here we implement a framework to degrade accuracy error with the user in the loop. We then approximate calibration error, with those degradations applied in each block to provide an “Effective Gaze Error.” Participants selected single targets (3° or 5° diameter) using an eye tracking sampling frequency and display rate of 120 Hz. Higher “Effective Gaze Error” on smaller targets resulted in decreased human performance and subjective evaluations. Our experiment framework and results provide a starting point for future studies assessing how gaze accuracy degradation impacts performance, beyond interactions tasks.

Prior research states that frequencies below 75 Hz in eye-tracking data represent the primary eye movement termed “signal” while those above 75 Hz are deemed “noise”. This study examines the biometric significance of this signal-noise distinction and its privacy implications. There are important individual differences in a person’s eye movement, which lead to reliable biometric performance in the “signal” part. Despite minimal eye-movement information in the “noise” recordings, there might be significant individual differences. Our results confirm the “signal” predominantly contains identity-specific information, yet the “noise” also possesses unexpected identity-specific data. This consistency holds for both short-(≈ 20 min) and long-term (≈ 1 year) biometric evaluations. Understanding the location of identity data within the eye movement spectrum is essential for privacy preservation.

This paper performs the crucial work of establishing a baseline for gaze-driven authentication performance to begin answering fundamental research questions using a very large dataset of gaze recordings from 9202 people with a level of eye tracking (ET) signal quality equivalent to modern consumer-facing virtual reality (VR) platforms. The size of the employed dataset is at least an order-of-magnitude larger than any other dataset from previous related work. Binocular estimates of the optical and visual axes of the eyes and a minimum duration for enrollment and verification are required for our model to achieve a false rejection rate (FRR) of below 3% at a false acceptance rate (FAR) of 1 in 50,000. In terms of identification accuracy which decreases with gallery size, we estimate that our model would fall below chance-level accuracy for gallery sizes of 148,000 or more. Our major findings indicate that gaze authentication can be as accurate as required by the FIDO standard when driven by a state-of-the-art machine learning architecture and a sufficiently large training dataset.

Thanks to the eye-tracking sensors that are embedded in emerging consumer devices like the Vive Pro Eye, we demonstrate that it is feasible to deliver user authentication via eye movement biometrics.

The power requirements of video-oculography systems can be prohibitive for high-speed operation on portable devices. Recently, low-power alternatives such as photosensors have been evaluated, providing gaze estimates at high frequency with a trade-off in accuracy and robustness. Potentially, an approach combining slow/high-fidelity and fast/low-fidelity sensors should be able to exploit their complementarity to track fast eye motion accurately and robustly. To foster research on this topic, we introduce OpenSFEDS, a near-eye gaze estimation dataset containing approximately 2M synthetic camera-photosensor image pairs sampled at 500 Hz under varied appearance and camera position. We also formulate the task of sensor fusion for gaze estimation, proposing a deep learning framework consisting in appearance-based encoding and temporal eye-state dynamics. We evaluate several single- and multi-rate fusion baselines on OpenSFEDS, achieving 8.7% error decrease when tracking fast eye movements with a multi-rate approach vs. a gaze forecasting approach operating with a low-speed sensor alone.

The recent emergence of ubiquitous, multi-platform eye tracking has raised user privacy concerns involving a threat that we have termed “cross-platform identity linkage.” This privacy violation may occur when a person is re-identified across multiple eye tracking-enabled platforms using personally identifying information that is implicitly expressed through their eye movement. We present an empirical investigation quantifying a modern eye movement biometric model’s ability to link subject identities across three different eye tracking devices using eye movement signals from each device. We show that a state-of-the art eye movement biometrics model demonstrates above-chance levels of biometric performance (34.99% equal error rate, 15% rank-1 identification rate) when linking user identities across one pair of devices, but not for the other. Considering these findings, we also discuss the impact that eye tracking signal quality has on the model’s ability to meaningfully associate a subject’s identity between two substantially different eye tracking devices. Our investigation advances a fundamental understanding of the privacy risks for identity linkage across platforms by employing both quantitative and qualitative measures of biometric performance, including a visualization of the model’s ability to distinguish genuine and imposter authentication attempts across platforms.

This paper proposes a novel evaluation framework, termed "critical evaluation periods," for evaluating continuous gaze prediction models. This framework emphasizes prediction performance when it is most critical for gaze prediction to be accurate relative to user perception. Based on perceptual characteristics of the human visual system such as saccadic suppression, this framework provides a more practical assessment of gaze prediction performance for gaze-contingent rendering compared to the dominant sample-by-sample evaluation strategy employed in literature, which overemphasizes performance during easy-to-predict periods of fixation. Using a case study with a lightweight deep learning gaze prediction model, we observe a significant discrepancy in the reported prediction accuracy between the proposed critical evaluation periods and the dominant evaluation strategy employed in literature. Based on our findings, we suggest that the proposed framework is more suitable for evaluating the performance of continuous gaze prediction models intended for gaze-contingent rendering applications.

We present an analysis of the eye tracking signal quality of the HoloLens 2’s integrated eye tracker. Signal quality was measured from eye movement data captured during a random saccades task from a new eye movement dataset collected on 30 healthy adults. We characterize the eye tracking signal quality of the device in terms of spatial accuracy, spatial precision, temporal precision, linearity, and crosstalk. Most notably, our evaluation of spatial accuracy reveals that the eye movement data in our dataset appears to be uncalibrated. Recalibrating the data using a subset of our dataset task produces notably better eye tracking signal quality.

Comparing the performance of new eye-tracking devices against an established benchmark is vital for identifying differences in the way eye movements are reported by each device. This paper introduces a new paired data set comprised of eye movement recordings captured simultaneously with both the EyeLink 1000—considered the “gold standard” in eye-tracking research studies—and the recently released AdHawk MindLink eye tracker. Our work presents a methodology for simultaneous data collection and a comparison of the resulting eye-tracking signal quality achieved by each device.

The Oculomotor plant mathematical model (OPMM) is a dynamic system that describes a human eye in motion. In this study, we focus on an anatomically inspired homeomorphic model where every component is a mathematical representation of a certain biological phenomenon of a real oculomotor plant. This approach estimates internal state of oculomotor plant from recorded eye movements. In the past, the utility of such models was shown to be useful in biometrics and gaze contingent rendering via eye movement prediction. In previous studies, an implicit underlying assumption was that a set of parameters estimated for a certain subject should remain consistent in time and generalize to unseen data. We note a major drawback of the prior work, as it operated under this assumption without explicit validation. This work creates a quantifiable baseline for the specific OPMM where the generalizability of the model parameters is the foundational property of their estimation.

Iris-based biometric authentication is a wide-spread biometric modality due to its accuracy, among other benefits. Improving the resistance of iris biometrics to spoofing attacks is an important research topic. Eye tracking and iris recognition devices have similar hardware that consists of a source of infra-red light and an image sensor. This similarity potentially enables eye tracking algorithms to run on iris-driven biometrics systems. The present work advances the state-of-the-art of detecting iris print attacks, wherein an imposter presents a printout of an authentic user’s iris to a biometrics system. The detection of iris print attacks is accomplished via analysis of the captured eye movement signal with a deep learning model. Results indicate better performance of the selected approach than the previous state-of-the-art.

Eye movement biometrics (EMB) is a relatively recent behavioral biometric modality that may have the potential to become the primary authentication method in virtual- and augmented-reality (VR/AR) devices due to their emerging use of eye-tracking sensors to enable foveated rendering techniques. However, existing EMB models have yet to demonstrate levels of performance that would be acceptable for real-world use. The present study proposes an improved methodology for EMB with the goal of satisfying the FIDO Biometrics Requirements’ recommendation of 5% false rejection rate at 1-in-10,000 false acceptance rate. A DenseNet-based convolutional neural network is proposed that is memory-efficient, relatively quick to train, and has only ~123K learnable parameters. The model is trained over an array of different eye-tracking tasks to improve the generalizability of learned features. Authentication performance is evaluated on a held-out set of up to 59 individuals across different eye-tracking tasks, test-retest intervals, and with increasing amounts of data available for enrollment and authentication. The impact of degraded sampling rates and spatial precision on authentication performance is also briefly explored to set the stage for future research targeting modern VR/AR devices. The proposed technique not only outperforms the previous state of the art but is also the first to approach a level of authentication performance that would be acceptable for real-world use.

We present a large scale data set of eye-images captured using a virtual-reality (VR) head mounted display mounted with two synchronized eye-facing cameras at a frame rate of 200 Hz under controlled illumination. This dataset is compiled from video capture of the eye-region collected from 152 individual participants and is divided into four subsets: (i) 12,759 images with pixel-level annotations for key eye-regions: iris, pupil and sclera (ii) 252,690 unlabeled eye-images, (iii) 91,200 frames from randomly selected video sequences of 1.5 seconds in duration, and (iv) 143 pairs of left and right point cloud data compiled from corneal topography of eye regions collected from a subset, 143 out of 152, participants in the study. A baseline experiment has been evaluated on the dataset for the task of semantic segmentation of pupil, iris, sclera and background, with the mean intersection-over-union (mIoU) of 98.3 %. We anticipate that this dataset will create opportunities to researchers in the eye tracking community and the broader machine learning and computer vision community to advance the state of eye-tracking for VR applications, which in its turn will have greater implications in Human-Computer Interaction.

Metric learning is a valuable technique for enabling the ongoing enrollment of new users within biometric systems. While this approach has been heavily employed for other biometric modalities such as facial recognition, applications to eye movements have only recently been explored. This manuscript further investigates the application of metric learning to eye movement biometrics. A set of three multilayer perceptron networks are trained for embedding feature vectors describing three classes of eye movements: fixations, saccades, and post-saccadic oscillations. The network is validated on a dataset containing eye movement traces of 269 subjects recorded during a reading task. The proposed algorithm is benchmarked against a previously introduced statistical biometric approach. While mean equal error rate (EER) was increased versus the benchmark method, the proposed technique demonstrated lower dispersion in EER across the four test folds considered herein.

Photosensor oculography (PSOG) is a promising solution for reducing the computational requirements of eye tracking sensors in wireless virtual and augmented reality platforms. This paper proposes a novel machine learning-based solution for addressing the known performance degradation of PSOG devices in the presence of sensor shifts. Namely, we introduce a convolutional neural network model capable of providing shift-robust end-to-end gaze estimates from the PSOG array output. Moreover, we propose a transfer-learning strategy for reducing model training time. Using a simulated workflow with improved realism, we show that the proposed convolutional model offers improved accuracy over a previously considered multilayer perceptron approach. In addition, we demonstrate that the transfer of initialization weights from pre-trained models can substantially reduce training time for new users. In the end, we provide the discussion regarding the design trade-offs between accuracy, training time, and power consumption among the considered models.

The importance of normalizing biometric features or matching scores is understood in the multimodal biometric case, but there is less attention to the unimodal case. Prior reports assess the effectiveness of normalization directly on biometric performance. We propose that this process is logically comprised of two independent steps: (1) methods to equalize the effect of each biometric feature on the similarity scores calculated from all the features together and (2) methods of weighting the normalized features to optimize biometric performance. In this report, we address step 1 only and focus exclusively on normally distributed features. We show how differences in the variance of features lead to differences in the strength of the influence of each feature on the similarity scores produced from all the features. Since these differences in variance have nothing to do with importance in the biometric sense, it makes no sense to allow them to have greater weight in the assessment of biometric performance. We employed two types of features: (1) real eye-movement features and (2) synthetic features. We compare six variance normalization methods (histogram equalization, L1-normalization, median normalization, z-score normalization, min-max normalization, and L-infinite normalization) and one distance metric (Mahalanobis distance) in terms of how well they reduce the impact of the variance differences. The effectiveness of different techniques on real data depended on the strength of the inter-correlation of the features. For weakly correlated real features and synthetic features, histogram equalization was the best method followed by L1 normalization.

As eye tracking can reduce the computational burden of virtual reality devices through a technique known as foveated rendering, we believe not only that eye tracking will be implemented in all virtual reality devices, but that eye tracking biometrics will become the standard method of authentication in virtual reality. Thus, we have created a real-time eye movement-driven authentication system for virtual reality devices. In this work, we describe the architecture of the system and provide a specific implementation that is done using the FOVE head-mounted display. We end with an exploration into future topics of research to spur thought and discussion.

Virtual reality (VR) is employed in a variety of different applications. It is our belief that eye-tracking is going to be a part of the majority of VR devices that will reduce computational burden via a technique called foveated rendering and will increase the immersion of the VR environment. A promising technique to achieve low energy, fast, and accurate eye-tracking is photo-sensor oculography (PS-OG). PS-OG technology enables tracking a user's gaze location at very fast rates - 1000Hz or more, and is expected to consume several orders of magnitude less power compared to a traditional video-oculography approach. In this demo we present a prototype of a PS-OG system that we started to develop. The long-term aim of our project is to develop a PS-OG system that is robust to sensor shifts. As a first step we have built a prototype that allows us to test different sensors and their configurations, as well as record and analyze eye-movement data.

This paper presents a renewed overview of photosensor oculography (PSOG), an eye-tracking technique based on the use of simple photosensors (usually infrared) to measure the overall amount of reflected light while the eye rotates. PSOG can provide high spatial and temporal resolution, low tracking latency, and reduced power consumption. The deeper examination of this technique is particularly important, given the current needs for efficient eye-tracking mechanisms by the emerging interaction devices, e.g., augmented and virtual reality headsets. In our investigation, we employ an adjustable simulation framework to examine the eye-tracking performance when changing the parameters of different PSOG designs. We focus our examination on eye-tracking accuracy and crosstalk, two crucial characteristics for the seamless interaction via eye-tracking. Our experiments demonstrate the parameter values (or ranges) that lead to different levels of performance for accuracy and crosstalk. We also show that the optimization of accuracy and crosstalk is often driven by competing conditions, and thus, we perform a multi-objective optimization procedure and estimate the respective trade-off parameters. Finally, we present the effects from sensor shifts and evaluate the resulting increase in accuracy error. Our results and analysis can be used to facilitate the selection of parameters in future PSOG systems.

This document describes a method for detecting the onset of eye fatigue and how it could be implemented in an existing live framework. The proposed method, which uses fixation data, does not rely as heavily on the sampling rate of the eye tracker as do methods which use saccade data, making it more suitable for lower cost eye trackers such as mobile and wearable devices. By being able to detect eye fatigue with such eye trackers, it becomes possible to react to the development of fatigue in virtually any environment, such as by alerting drivers that they appear fatigued and may want to pull over. It could also be used to aid in developing interfaces that are more user-friendly by noting at which point a user becomes fatigued while navigating the interface.

Previous research shows that human eye movements can serve as a valuable source of information about the structural elements of the oculomotor system and they also can open a window to the neural functions and cognitive mechanisms related to visual attention and perception. The research field of eye movement-driven biometrics explores the extraction of individual-specific characteristics from eye movements and their employment for recognition purposes. In this work, we present a study for the incorporation of dynamic saccadic features into a model of eye movement-driven biometrics. We show that when these features are added to our previous biometric framework and tested on a large database of 322 subjects, the biometric accuracy presents a relative improvement in the range of 31.6--33.5% for the verification scenario, and in range of 22.3--53.1% for the identification scenario. More importantly, this improvement is demonstrated for different types of visual stimulus (random dot, text, video), indicating the enhanced robustness offered by the incorporation of saccadic vigor and acceleration cues.

This paper presents an objective evaluation of the effects of environmental factors, such as stimulus presentation and eye tracking specifications, on the biometric accuracy of oculomotor plant characteristic biometrics. This paper examines the largest known data set for eye movement biometrics, with eye movements recorded from 323 subjects over multiple sessions. Six spatial precision tiers (0.01°, 0.11°, 0.21°, 0.31°, 0.41°, and 0.51°), six temporal resolution tiers (1000, 500, 250, 120, 75, and 30 Hz), and three stimulus types (horizontal, random, and textual) are evaluated to identify acceptable conditions under which to collect eye movement data. The results suggest the use of eye tracking equipment providing at least 0.1° spatial precision and 30-Hz sampling rate for biometric purposes, and the use of a horizontal pattern stimulus when using the 2-D oculomotor plant model developed by Komogortsev et al.

This paper investigates liveness detection techniques in the area of eye movement biometrics. We investigate a specific scenario, in which an impostor constructs an artificial replica of the human eye. Two attack scenarios are considered: 1) the impostor does not have access to the biometric templates representing authentic users, and instead utilizes average anatomical values from the relevant literature and 2) the impostor gains access to the complete biometric database, and is able to employ exact anatomical values for each individual. In this paper, liveness detection is performed at the feature and match score levels for several existing forms of eye movement biometric, based on different aspects of the human visual system. The ability of each technique to differentiate between live and artificial recordings is measured by its corresponding false spoof acceptance rate, false live rejection rate, and classification rate. The results suggest that eye movement biometrics are highly resistant to circumvention by artificial recordings when liveness detection is performed at the feature level. Unfortunately, not all techniques provide feature vectors that are suitable for liveness detection at the feature level. At the match score level, the accuracy of liveness detection depends highly on the biometric techniques employed.

This document describes the software framework of an ocular biometric system. The framework encompasses several interconnected components that allow an end-user to perform biometric enrollment, verification, and identification with most common eye tracking devices. The framework, written in C#, includes multiple state-of-the-art biometric algorithms and information fusion techniques, and can be easily extended to utilize new biometric techniques and eye tracking devices.

This work investigates the possibility of detecting iris print-attacks via the analysis of a number of gaze-related features acquired in a process of eye tracking. Gaze estimation algorithms employ models based on the physical structure and function of the eye, providing thus a number of salient features that can be potentially employed for the detection of spoofing print-attacks. In our study, a combined dataset was assembled for the investigation of these features, consisting of eye movement recordings and the corresponding iris images collected from 100 subjects. The collected iris images were utilized in direct implementation of iris print-attacks against an eye tracking device. We developed a methodology for the detection of spoof indicative artifacts in the recorded signals, and fed the extracted features from the live and spoof eye signals into a two-class SVM classifier. The obtained results indicate a best correct classification rate (CCR) of 95.7%. Furthermore, we demonstrate the moderate decrease in liveness detection rates during subsampling of the eye movement signal to frequencies as low as 15 Hz. This result indicates the usefulness of running gaze estimation algorithms on existing iris recognition devices where such sampling frequency rate is common.

This article proposes and evaluates a novel biometric approach utilizing the internal, nonvisible, anatomical structure of the human eye. The proposed method estimates the anatomical properties of the human oculomotor plant from the measurable properties of human eye movements, utilizing a two-dimensional linear homeomorphic model of the oculomotor plant. The derived properties are evaluated within a biometric framework to determine their efficacy in both verification and identification scenarios. The results suggest that the physical properties derived from the oculomotor plant model are capable of achieving 20.3% equal error rate and 65.7% rank-1 identification rate on high-resolution equipment involving 32 subjects, with biometric samples taken over four recording sessions; or 22.2% equal error rate and 12.6% rank-1 identification rate on low-resolution equipment involving 172 subjects, with biometric samples taken over two recording sessions.

This paper proposes a method for the extraction of biometric features from the spatial patterns formed by eye movements during an inspection of dynamic visual stimulus. In the suggested framework, each eye movement signal is transformed into a time-constrained decomposition by using a probabilistic representation of spatial and temporal features related to eye fixations and called fixation density map (FDM). The results for a large collection of eye movements recorded from 200 individuals indicate the best equal error rate of 10.8% and Rank-1 identification rate as high as 51%, which is a significant improvement over existing eye movement-driven biometric methods. In addition, our experiments reveal that a person recognition approach based on the FDM performs well even in cases when eye movement data are captured at lower than optimum sampling frequencies. This property is very important for the future ocular biometric systems where existing iris recognition devices could be employed to combine eye movement traits with iris information for increased security and accuracy. Considering that commercial iris recognition devices are able to implement eye image sampling usually at a relatively low rate, the ability to perform eye movement-driven biometrics at such rates is of great significance.

MAGIC pointing techniques combine eye tracking with manual input. Since the mouse performs exceptionally well in a desktop setting, previous research on MAGIC pointing either resulted in minor improvements, or the techniques were applied to alternative devices or environments. We design Animated MAGIC, a novel, target-agnostic MAGIC pointing technique, for the specific goal of beating the mouse in a desktop setting. To improve the eye-tracking accuracy, we develop a dynamic local calibration method that uses each selection as a local calibration point. We compare Animated MAGIC to mouse-only and Conservative MAGIC, one of the two original MAGIC pointing methods, in a Fitts' Law experiment. We conduct a user questionnaire to evaluate the usability of the interaction methods. Results suggest that Dynamic Local Calibration improves eye-tracking accuracy and, consequently, MAGIC pointing performance. Powered with Dynamic Local Calibration, Animated MAGIC outperformed mouse-only by 8% in terms of throughput. Both MAGIC pointing methods reduced the amount of hand movement by more than half.

This article assesses the ability of a two-dimensional (2D) linear homeomorphic oculomotor plant mathematical model to simulate normal human saccades on a 2D plane. The proposed model is driven by a simplified pulse-step neuronal control signal and makes use of linear simplifications to account for the unique characteristics of the eye globe and the extraocular muscles responsible for horizontal and vertical eye movement. The linear nature of the model sacrifices some anatomical accuracy for computational speed and analytic tractability, and may be implemented as two one-dimensional models for parallel signal simulation. Practical applications of the model might include improved noise reduction and signal recovery facilities for eye tracking systems, additional metrics from which to determine user effort during usability testing, and enhanced security in biometric identification systems. The results indicate that the model is capable of produce oblique saccades with properties resembling those of normal human saccades and is capable of deriving muscle constants that are viable as biometric indicators. Therefore, we conclude that sacrifice in the anatomical accuracy of the model produces negligible effects on the accuracy of saccadic simulation on a 2D plane and may provide a usable model for applications in computer science, human-computer interaction, and related fields.

This paper presents an objective evaluation of the effects of eye tracking specification and stimulus presentation on the biometric viability of complex eye movement patterns. Six spatial accuracy tiers (0.5°, 1.0°, 1.5°, 2.0°, 2.5°, 3.0°), six temporal resolution tiers (1000, 500, 250, 120, 75, 30 Hz), and five stimulus types (simple, complex, cognitive, textual, random) are evaluated to identify acceptable conditions under which to collect eye movement data. The results suggest the use of eye tracking equipment capable of at least 0.5°spatial accuracy and 250 Hz temporal resolution for biometric purposes, whereas stimulus had little effect on the biometric viability of eye movements.

Automated detection of excessive visual search (ES) experienced by a user during software use presents the potential for substantial improvement in the efficiency of supervised usability analysis. This paper presents an objective evaluation of several methods for the automated segmentation and classification of ES intervals from an eye movement recording, a technique that can be utilized to aid in the identification of usability problems during software usability testing. Techniques considered for automated segmentation of the eye movement recording into unique intervals include mouse/keyboard events and eye movement scanpaths. ES is identified by a number of eye movement metrics, including: fixation count, saccade amplitude, convex hull area, scanpath inflections, scanpath length, and scanpath duration. The ES intervals identified by each algorithm are compared to those produced by manual classification to verify the accuracy, precision, and performance of each algorithm. The results indicate that automated classification can be successfully employed to substantially reduce the amount of recorded data reviewed by HCI experts during usability testing, with relatively little loss in accuracy.

This work describes the design and implementation of an eye tracking system on an unmodified common tablet PC. A neural network eye tracker is employed as a solution to eye tracking in the visible spectrum of light. We discuss the challenges related to image recognition and processing, and provide an objective evaluation of the accuracy and sampling rate of eye-gaze-based interaction with such an eye tracker. The results indicate that it is possible to obtain an average accuracy of 4.42° and a sampling rate of 0.70 Hz with the described system.

This paper presents an objective evaluation of various eye movement-based biometric features and their ability to accurately and precisely distinguish unique individuals. Eye movements are uniquely counterfeit resistant due to the complex neurological interactions and the extraocular muscle properties involved in their generation. Considered biometric candidates cover a number of basic eye movements and their aggregated scanpath characteristics, including: fixation count, average fixation duration, average saccade amplitudes, average saccade velocities, average saccade peak velocities, the velocity waveform, scanpath length, scanpath area, regions of interest, scanpath inflections, the amplitude-duration relationship, the main sequence relationship, and the pairwise distance between fixations. As well, an information fusion method for combining these metrics into a single identification algorithm is presented. With limited testing this method was able to identify subjects with an equal error rate of 27%. These results indicate that scanpath-based biometric identification holds promise as a behavioral biometric technique.

There has been increased interest in reliable, non-intrusive methods of biometric identification due to the growing emphasis on security and increasing prevalence of identity theft. This paper presents a new biometric approach that involves an estimation of the unique oculomotor plant (OP) or eye globe muscle parameters from an eye movement trace. These parameters model individual properties of the human eye, including neuronal control signal, series elasticity, length tension, force velocity, and active tension. These properties can be estimated for each extraocular muscle, and have been shown to differ between individuals. We describe the algorithms used in our approach and the results of an experiment with 41 human subjects tracking a jumping dot on a screen. Our results show improvement over existing eye movement biometric identification methods. The technique of using Oculomotor Plant Mathematical Model (OPMM) parameters to model the individual eye provides a number of advantages for biometric identification: it includes both behavioral and physiological human attributes, is difficult to counterfeit, non-intrusive, and could easily be incorporated into existing biometric systems to provide an extra layer of security.

This paper presents a set of qualitative and quantitative scores designed to assess performance of any eye movement classification algorithm. The scores are designed to provide a foundation for the eye tracking researchers to communicate about the performance validity of various eye movement classification algorithms. The paper concentrates on the five algorithms in particular: Velocity Threshold Identification (I-VT), Dispersion Threshold Identification (I-DT), Minimum Spanning Tree Identification (MST), Hidden Markov Model Identification (I-HMM) and Kalman Filter Identification (I-KF). The paper presents an evaluation of the classification performance of each algorithm in the case when values of the input parameters are varied. Advantages provided by the new scores are discussed. Discussion on what is the "best" classification algorithm is provided for several applications. General recommendations for the selection of the input parameters for each algorithm are provided.

Usability testing activities have numerous benefits in theory, yet they are often overlooked or disregarded in practice. A testing paradigm which yields objective, quantitative results would likely lead to more widespread adoption of usability evaluation activities. Total-effort metrics is such a novel framework. This paper describes a usability study conducted using a total-effort metrics approach. In this study, subjects interact with three interfaces which have varying element layout proximities. The time and effort measures of time-on-task, total keystrokes, correctional keystrokes, saccade amplitude (point-to-point eye movement) and gaze-path traversal are recorded and analyzed. The findings of the study demonstrate a correlation between the intrinsic effort of an interface and its usability as predicted by extant interface layout guidelines.

Designing Human Computer Interfaces is one of the more important and difficult design tasks. The tools for verifying the quality of the interface are frequently expensive or provide feedback too far after the design of the interface as to make it meaningless. To improve the interface usability, designers need a verification tool providing immediate feedback at a low cost. Using an effort-based measure of usability, it is possible for a designer to estimate the effort a subject might expend to complete a specific task. In this paper, we develop the notion of designer's effort for evaluating interface usability for new designs and Commercial-Off-The-Shelf software. Designer's effort provides a technique to evaluate human interface before completing the development of the software and provides feedback from usability tests conducted using the effort-based evaluation technique.

This paper builds a two dimensional oculomotor plant mathematical model (2DOPMM) that is capable of generating eye movement trace on a two dimensional plane. The key difference between the proposed model and the models presented previously is a design that is geared towards linearity and capability of integration into a real-time human computer interaction system while providing force output for each extraocular muscle with values close to physiological measurements. The model is represented as a twelve order system created by a set of linear mechanical components representing major anatomical properties of extraocular muscles and the eye globe: muscle location, elasticity, viscosity, eye-globe rotational inertia, muscle active state tension, length tension and force velocity relationships. Linearity is a key point ensuring a real-time performance in an online implementation of the model with twelve order representation providing close match to the eye anatomical structure. Practical applications of the proposed model lie in the area of extraocular muscle effort estimation and human computer interaction.

The goal of this paper is to predict future horizontal eye movement trajectories within a specified time interval. To achieve this goal a linear horizontal oculomotor plant mechanical model is developed. The model consists of the eye globe and two extraocular muscles: lateral and medial recti. The model accounts for such anatomical properties of the eye as muscle location, elasticity, viscosity, eye-globe rotational inertia, muscle active state tension, length tension and force velocity relationships. The mathematical equations describing the oculomotor plant mechanical model are transformed into a Kalman filter form. Such transformation provides continuous eye movement prediction with a high degree of accuracy. The model was tested with 21 subjects and three multimedia files. Practical application of this model lies with direct eye gaze input and interactive displays systems as a method to compensate for detection, transmission and processing delays.

This article designs a real-time perceptual compression system (RTPCS) based on eye-gaze-position analysis. Our results indicate that the eye-gaze-position containment metric provides more efficient and effective evaluation of an RTPCS than the eye fixation containment. The presented RTPCS is designed for a network communication scenario with a feedback loop delay. The proposed RTPCS uses human visual system properties to compensate for the delay and to provide high ratios of multimedia compression.

Human eyes have limited perception capabilities. Only 2 degrees of our 180 degree vision field provide the highest quality of perception. Due to this fact the idea of perceptual attention focus emerged to allow a visual content to be changed in a way that only part of the visual field where a human attention is directed to is encoded with a high quality. The image quality in the periphery can be reduced without a viewer noticing it. This compression approach allows a significant decrease in bit-rate for a video stream, and in the case of the 3D stream rendering, it decreases the computational burden. A number of previous researchers have investigated the topic of real-time perceptual attention focus but only for a single viewer. In this paper we investigate a dynamically changing multi-viewer scenario. In this type of scenario a number of people are watching the same visual content at the same time. Each person is using eye-tracking equipment. The visual content (video, 3D stream) is sent through a network with a large transmission delay. The area of the perceptual attention focus is predicted for the viewers to compensate for the delay value and identify the area of the image which requires highest quality coding.

Human eyes have limited perception capabilities; for example, only 2 degrees of our 140 degree vision field provide the highest quality of perception. Due to this fact the idea of perceptual focus emerged to allow a visual content to be changed in a way that only part of the visual field where a human gaze is directed is encoded with a high quality. The image quality in the periphery can be reduced without a viewer noticing it. This compression approach allows a significant decrease in the number of bits required for image encoding, and in the case of the 3D image rendering, it decreases the computational burden. A number of previous researchers have investigated the topic of perceptual focus but only for a single viewer. In our research we investigate a dynamically changing multi-viewer scenario. In this type of scenario a number of people are watching the same visual content at the same time. Each person has his/her own perceptual focus area which changes over time. The visual content is sent through a network with a fixed delay/lag which provides an additional challenge to the whole scheme. The goal of our work was to investigate and develop a method of multi-viewer perceptual focus zones adaptation for real-time media perceptual compression and transmission. In our research we also look into the impact that such a method can have on transmission bandwidth and computational burden reduction.

No abstract available.

Approximately 2 degrees in our 140 degree vision span has sharp vision. Many researchers have been fascinated by the idea of eye-tracking integrated perceptual compression of an image or video, yet any practical system has yet to emerge. The unique challenge presented by real time perceptual video streaming is how to handle the fast nature of the human eye and provide its integration with computationally intensive video transcoding scheme. The delay introduced by video transmission in the network presents a difficulty. This delay creates a problem when we try to use information about eye movements for perceptual encoding. In this paper we discuss a new approach to the eye-tracker based video compression. Rather than relying on the point of gaze, this novel scheme tracks a vicinity of interest and offers a prediction mechanism for eye movements. The described system compensates the interim eye movements between the sampling and actual coding. The proposed scheme can be applied to a large variety of today's video compression standards. We have developed an eye gaze-aware MPEG-2 transcoder that can perceptually re-encode a live video stream in real time. The experiments we have conducted illustrate the substantial impact this integrated prediction method has on perceptual video compression and bit-rate reduction.

The possibility of perceptual compression using live eye-tracking has been anticipated for some time by many researchers. Among the challenges of real-time eye-gaze based perceptual video compression is how to handle the fast nature of eye movements with a relative complexity of video transcoding and also take into the account a delay associated with transmission in the network. Such delay requires an additional consideration in perceptual encoding because it increases the size of the area that requires high quality coding. In this paper we present a hybrid scheme, one of the first to our knowledge, which combines eye-tracking with fast in-line scene analysis to drastically narrow down the high acuity area without the loss of eye-gaze containment.

In this paper we discuss architecture for network aware adaptive systems for next generation networks. We present in the context of a novel cognizant video transcoding system, which is capable of negotiating local network state based rate and let the video propagate over extreme network with highly asymmetric link and node capacities utilizing knowlege about the network, content protocol and the content itself.