The application of chaos theory in motion data analysis is the subject of the paper. The main novelty is related to the proposal of a feature extraction technique based on transfer learning and focused on the properties that quantify the level of chaos of non-linear dynamical models. The introduced method is successfully examined in the gait-based human sex recognition challenge. There are two key stages of the proposed approach. In the first one Long Short-Term Memory (LSTM) networks are trained to recognize chaotic and non-chaotic systems based on their univariate observation sequences. They are built using state variables of well-known non-linear dynamical models. Afterward, the network extracts interpretable features of every pose parameter of the human skeleton model. Thus, multivariate motion sequences are decomposed into univariate ones and then processed separately by the LSTM network. As a result, motion descriptors are obtained. Their features are assessed for significant differences between the populations of females and males. Moreover, the performance of the supervised classification is evaluated. In numerical experiments, highly precise motion capture measurements, registered by the gold-standard Vicon system, are used. The obtained results confirm that the proposed approach provides robust and discriminative motion features. They allow for efficient distinction of gait performed by females and males with more than 92% accuracy. Moreover, as statistically significant differences are found and the areas under the receiver operating curve are computed for feature values, it is feasible to assess joints whose movements differ most strongly.

Contains fulltext : 326363.pdf (Publisher’s version ) (Open Access)

Post-saccadic oscillations (PSOs) are commonly observed in eye movement data but are often neglected, resulting in misclassifications of eye movement events and inaccurate interpretations. This study utilizes an event detection model based on CNN architecture that classifies eye movement data into three categories of events: fixations, saccades, and PSOs. The model is used to classify eye movement data collected during text reading. The results of the model are then compared with the commonly used classic SR Research software algorithm, which classifies eye movement data into fixations and saccades only. We compare the results and show that PSO detection significantly influences statistics for fixations (like average duration) that are commonly used to analyze different properties of the text being read (like complexity of the text). We postulate that PSOs should be taken into account in future eye movements when reading research.

The quality of welds is a critical factor in ensuring the structural integrity and safety of manufactured components across industries such as aerospace, automotive, and construction. Traditional welding inspection methods often rely on manual or semiautomated techniques, which can be time consuming, subjective, and prone to human error. This paper explores the application of artificial intelligence (AI), specifically deep learning models, in real-time monitoring and evaluation of welding quality. By leveraging recurrent neural networks (RNNs), the proposed approach analyzes data from various sensors to detect defects. Deep learning models are trained on a dataset of 11 labeled sensor signals, allowing the system to learn complex patterns associated with both high- and low-quality welds. Experimental results demonstrate that the AI-based monitoring system can be used to automatically detect welding correctness. The integration of deep learning in welding quality control is a step towards fully automated and intelligent manufacturing processes. Furthermore, the implementation of this technology offers significant economic benefits by reducing inspection costs, minimizing rework and scrap rates, and improving overall production efficiency.

Dyslexia is a serious problem for society. Not-diagnosed dyslexia may lead to problems with comprehension and result in low self-esteem. Diagnosing dyslexia is difficult for adults as they develop different ways to mask their impairment. One of the possible ways to detect dyslexia is to analyze people’s eye movements while they are reading. Such methods typically convert raw eye movements to events - fixations and saccades - and then try to map fixations to words of the read text. Different statistics are used, like average fixations’ duration, number of refixations, and so on. This paper proposes a technique that diagnoses dyslexia based on raw eye movements without any preprocessing. This method is thus much more straightforward and does not require any parameters to be used to detect fixations and map them to words. Our method converts raw eye movement recordings to Gaze Self Similarity Plots and then trains the Convolutional Neural Network with these plots as input images. The method exhibits a state-of-the-art performance on a public dataset of adult Danish readers with accuracy equal to 89% and AUC equal to 0.93.

The paper briefly describes the main assumptions and preliminary results of an ongoing project that aims to propose objective methods for assessing the mental state of the patient with locked-in syndrome. The primary assumption of the project is to utilize eye movements as it is frequently the only way to communicate with such patients. During the project, specially designed cognitive tests were prepared and conducted with - to date - over 50 patients with different states of consciousness. As the project is still unfinished, the paper only reports some assumptions and a few demographic data.

Various letter arrangements were analysed for the virtual Side keyboard developed during earlier studies. The key layout applied there, based on the users’ opinions, could have been more convenient. Therefore, two additional solutions were introduced and verified. They were alphabetical and QWERTY-like letter layouts. Twenty participants were involved in the experiment, and the task was to type a sentence with their eyes. One sentence was used for all participants and all keyboards. Several metrics were studied: typing time, speed, and number of mistakes. The obtained results confirmed the previously collected opinions, favoring the alphabetical and QWERTY-like letter arrangements. It was especially visible for people without eye-tracking experience.

The paper presents a deep-learning model that may be used to calculate saliency maps for video content. Classic saliency algorithms take into account only spatial information obtained from an image to generate a saliency map showing regions of higher importance—hence, regions where people are more likely to turn their gaze and attention. Algorithms for video stimuli add temporal information about the movements of objects on a frame-to-frame basis, resulting in more complex three-dimensional architectures. The paper analyses existing models and proposes a model based on one of them. The model’s performance is compared with the literature using four widely accepted measures (AUC, NSS, SIM, and CC). It is comparable and, in many cases, even better than already published models. Additionally, because of some improvements in the architecture, it is significantly faster in terms of the number of frames processed per second.

We introduce pymovements: a Python package for analyzing eye-tracking data that follows best practices in software development, including rigorous testing and adherence to coding standards. The package provides functionality for key processes along the entire preprocessing pipeline. This includes parsing of eye tracker data files, transforming positional data into velocity data, detecting gaze events like saccades and fixations, computing event properties like saccade amplitude and fixational dispersion and visualizing data and results with several types of plotting methods. Moreover, pymovements also provides an easily accessible interface for downloading and processing publicly available datasets. Additionally, we emphasize how rigorous testing in scientific software packages is critical to the reproducibility and transparency of research, enabling other researchers to verify and build upon previous findings.

The presented studies investigated the possibility of eye movement usage for entering text using on-screen keyboards. The attention was focused on the keyboard layout and the time of typing a character. Three dwell times were used to select keyboard keys: 2-second, 1-second, and adaptive. The last of them required the development of an appropriate algorithm to adjust the character choice to the user's skills and to facilitate the use of the software. The paper analyzes whether the keyboard type and character selection time affect the text input speed and the number of errors made. During the experiment, in which seven people participated, the text consisting of approximately 70 characters was entered. This operation was repeated for each keyboard layout and dwell time. The studies have confirmed the reduced error number and the greater comfort for the adaptive selection solution.

It is common in eye movement acquisition that data comes from different devices and exhibits different sampling rates. One of the solutions to this problem is to recalculate the signal into another sampling rate. When upsampling is performed on a sequence of samples, it approximates the sequence that would have been obtained by sampling the signal at a higher rate. Many methods may be used for eye movement signal upsampling. Recently, Convolutional Neural Networks proved to be very efficient in upsampling (or supersampling) digital images. This paper attempts to utilize Convolutional Neural Networks using two architectures with different types of layers on the signal sampled with 125 Hz to obtain an eight-time increase in the sampling rate and produce the signal with a 1000 Hz sampling rate. The experiments on the GazeBase dataset proved that this solution is feasible, and the Convolutional Neural Network can learn the characteristic of the eye movement signal.

A novel approach for eye-typing using an on-screen keyboard was proposed in the paper. It was compared with the two applied in the previous studies. All utilized the dwell-time selection for writing letters, set to 1000 ms. This value was chosen due to the planned group of participants. Among 14 engaged people, there were four over the age of 50 and one aged 47. The text utilized in the research was prepared in two languages – Polish and English. The obtained results showed that the language used had no influence on typing efficiency. Additionally, they revealed the new keyboard layout was more convenient than one of the two chosen for comparison and similar to the second. Such conclusions were based on the experiment duration and the number of errors made. They were confirmed by the users’ opinions.

The paper focuses on utilizing eye movements for controlling computer programs. Four different approaches were considered, including gazing at an element, blinking, and two solutions for eye gestures. The methods were implemented in a simple photo viewer. However, the developed functionalities can be utilized in other applications (switching e-book pages or on-screen keyboard elements). Various sizes for gazed components were verified, as well as the time required for performing particular actions, the number of unwanted element choices, and needed gesture repetitions. The methods were also assessed by the participants of the experiments. Based on the obtained results, it may be reasoned that components of size 200px are convenient for the application control. The users’ opinions and the experiments’ outcomes revealed that the gaze-based method and gestures based on joining points are satisfactory solutions.

During eye-tracking experiments, a large amount of data is collected. Usually, it is stored in text files, or file formats meant for scientific data. Because these formats provide limited sets of mechanisms for more comprehensive data analysis, various database models were considered as an alternative for them in this study. They included a relational database defined in MS SQL Server and two commonly used NoSQL solutions – MongoDB and Cassandra. The data structures defined for each of them and their performance were compared based on a chosen set of operations. A special multi-database platform was developed, enabling access to all database models without knowing their mechanisms for manipulating data. The studies revealed that Cassandra is the best database to store eye-tracking data when adjusted to the data retrieving operations. MongoDB databases featured with similar performance for various query types, yet with worse execution time. The worst results were obtained for the relational database. Designed to avoid redundancy by assumption, it stores data in many tables, and many join operations are needed while reading data. This fact entails weaker operations efficiency.

The aim of the research is the introduction of new techniques that enable a visual comparison of scan-paths. Every eye tracking experiment produces many scan-paths, and one of the main challenges of eye tracking analysis is how two compare these scan-paths. A classic solution is to extract easily measurable features such as fixation durations or saccade lengths. There are also many more sophisticated techniques that compare two scan-paths using only spatial or both spatial and temporal information. These techniques typically return a value (or several values) that may be used as scan-path similarity/distance measure. However, there is still a lack of widely adopted methods that offer not only the measure but enable a visual comparison of scan-paths. The paper introduces two possible options: the Mutual Distance Plot for two scan-paths and the Warped Time Distance chart for the comparison of the theoretically unlimited number of scan-paths. It is shown that these visualizations may reveal information about relationships between two or more scan-paths on straightforward charts. The informativeness of the solution is analyzed using both artificial and real data.

With growing access to cheap low end eye trackers using simple web cameras, there is also a growing demand on easy and fast usage of this devices by untrained and unsupervised end users. For such users the necessity to calibrate the eye tracker prior to its first usage is often perceived as obtrusive and inconvenient. In the same time perfect accuracy is not necessary for many commercial applications. Therefore, the idea of implicit calibration attracts more and more attention. Algorithms for implicit calibration are able to calibrate the device without any active collaboration with users. Especially, a real time implicit calibration, that is able to calibrate a device on-the-fly, while a person uses an eye tracker, seems to be a reasonable solution to the aforementioned problems.

The aim of this research was to compare visual patterns while examining radiographs in groups of people with different levels and different types of expertise. Introducing the latter comparative base is the original contribution of these studies. The residents and specialists were trained in medical diagnosing of X-Rays and for these two groups it was possible to compare visual patterns between observers with different level of the same expertise type. On the other hand, the radiographers who took part in the examination - due to specific of their daily work - had experience in reading and evaluating X-Rays quality and were not trained in diagnosing. Involving this group created in our research the new opportunity to explore eye movements obtained when examining X-Ray for both medical diagnosing and quality assessment purposes, which may be treated as different types of expertise.

The paper describes the comparison of saliency maps calculated for a specific type of images to the ground truth fixations obtained in eye tracking experiment. The images being analyzed contained camouflaged animals. Because of the camouflage, it was expected that calculation of saliency would be the hard case. Several saliency maps were compared to the ground truth with a set of standard measures and a ranking based on their averaged performance was proposed. The results reveal that for the saliency map models, images of camouflaged animals are harder to analyze than ordinary ones, although, one model - Judd - appeared to work particularly well. Additionally, susceptibility of saliency maps to the perceived difficulty of the camouflage was discussed.

This paper presents the first attempt to fuse two different kinds of behavioral biometrics: mouse dynamics and eye movement biometrics. Mouse dynamics were collected without any special equipment, while an affordable The Eye Tribe eye tracker was used to gather eye movement data at a frequency of 30 Hz, which is also potentially possible using a common web camera. We showed that a fusion of these techniques is quite natural and it is easy to prepare an experiment that collects both traits simultaneously. Moreover, the fusion of information from both signals gave 6.8 % equal error rate and 92.9 % accuracy for relatively short registration time (20 s on average). Achieving such results were possible using dissimilarity matrices based on dynamic time warping distance.

The paper presents the algorithm supporting an implicit calibration of eye movement recordings. The algorithm does not require any explicit cooperation from users, yet it uses only information about a stimulus and an uncalibrated eye tracker output. On the basis of this data, probable fixation locations are calculated at first. Such a fixation set is used as an input to the genetic algorithm which task is to choose the most probable targets. Both information can serve to calibrate an eye tracker. The main advantage of the algorithm is that it is general enough to be used for almost any stimulation. It was confirmed by results obtained for a very dynamic stimulation which was a shooting game. Using the calibration function built by the algorithm it was possible to predict where a user will click with a mouse. The accuracy of the prediction was about 75%.

Dynamic development of high quality cameras and algorithms processing eye movement signals entails growing interests in using them in various areas of human-computer interaction. Determining subjects which user is looking at or controlling the operation of computer processes can serve as examples of these areas. However, making eye movement signal valuable requires some preparatory steps to be taken. They belong to a process called calibration aiming at creating a model for mapping output delivered by an eye tracker to user's gaze points. The quality of such model is assessed based on a calibration error defined as a difference between accurate data and this obtained from a model. The goal of the research presented in the paper was to analyse to what extent the calibration error depends on the specific participant's features - it is repeatable - or to what extent it may be avoided during the recalibration. Additionally an influence of two calibration method a polynomial and an artificial neural network (ANN) on the final results were studied as well.

The idea concerning usage of the eye movement for human identification has been known for 10 years. However, there is still lack of commonly accepted methods how to perform such identification. This paper describes the second edition of Eye Movement Verification and Identification Competition (EMVIC), which may be regarded as an attempt to provide some common basis for eye movement biometrics (EMB). The paper presents some details describing the organization of the competition, its results and formulates some conclusions for further development of EMB.

Eye movement is a new emerging modality in human computer interfaces. With better access to devices which are able to measure eye movements (so called eye trackers) it becomes accessible even in ordinary environments. However, the first problem that must be faced when working with eye movements is a correct mapping from an output of eye tracker to a gaze point – place where the user is looking at the screen. That is why the work must always be started with calibration of the device. The paper describes the process of calibration, analyses of the possible steps and ways how to simplify this process.