Recent advances on human motion analysis have made the extraction of skeleton structure feasible, even from single depth images. This has been proven quite informative for discriminating actions in a recognition scenario. In this context, we propose local descriptor that encodes relative position joint quadruples. Such coding implies similarity normalisation transform leads to compact (6D) view-invariant skeletal feature, referred as quad. Further, use Fisher kernel representation is...

This study scrutinizes the existing literature regarding use of augmented reality and gamification in education to establish its theoretical basis. A systematic review following Preferred Reporting Items for Systematic Reviews Meta-Analyses (PRISMA) statement was conducted. To provide complete valid information, all types related studies educational stages subjects throughout years were investigated. In total, 670 articles from 5 databases (Scopus, Web Science, Google Scholar, IEEE, ERIC)...

Vision Transformers (ViT) have shown rapid progress in computer vision tasks, achieving promising results on various benchmarks. However, due to the massive number of parameters and model design, \textit{e.g.}, attention mechanism, ViT-based models are generally times slower than lightweight convolutional networks. Therefore, deployment ViT for real-time applications is particularly challenging, especially resource-constrained hardware such as mobile devices. Recent efforts try reduce...

With the success of Vision Transformers (ViTs) in computer vision tasks, recent arts try to optimize performance and complexity ViTs enable efficient deployment on mobile devices. Multiple approaches are proposed accelerate attention mechanism, improve inefficient designs, or incorporate mobile-friendly lightweight convolutions form hybrid architectures. However, ViT its variants still have higher latency considerably more parameters than CNNs, even true for years-old MobileNet. In practice,...

This study aims to understand the public’s perspectives, sentiments, attitudes, and discourses regarding adoption, integration, use of augmented reality virtual in education general by analyzing social media data. Due its nature, Twitter was selected platform. Over 17 million tweets were retrieved from January 2010 December 2020 four datasets created. Two them referred these technologies two their educational use. The data analyzed using text mining, sentiment analysis (e.g., polarity...

This study aims to evaluate the impact of using augmented reality, gamification, and serious games in computer science education. The presents development process an educational mobile application, describes experiment that was conducted involved 117 higher education students, analyzes results a 49-item paper-based questionnaire. In total, 8 research questions were explored. revealed several benefits can be yielded when integrating such applications teaching learning activities actively...

This study aims to examine the combination of educational data mining and learning analytics with virtual reality, augmented mixed metaverse, its role in education, impact on teaching learning. Therefore, a systematic literature review, bibliometric scientific mapping analysis, content analysis are carried out based 70 relevant documents identified from six databases, namely, ACM, ERIC, IEEE, ScienceDirect, Scopus, Web Science (WoS) following PRISMA framework. The were separated into three...

Head-pose estimation has many applications, such as social event analysis, human-robot and human-computer interaction, driving assistance, so forth. is challenging because it must cope with changing illumination conditions, variabilities in face orientation appearance, partial occlusions of facial landmarks, well bounding-box-to-face alignment errors. We propose tu use a mixture linear regressions partially-latent output. This regression method learns to map high-dimensional feature vectors...

“Interaction” represents a critical term in the augmented and mixed reality ecosystem. Today, environments applications, interaction occupies joint space between any combination of humans, physical environment, computers. Although methods techniques have been extensively examined recent decades field human-computer interaction, they still should be reidentified context immersive realities. The latest technological advancements sensors, processing power technologies, including internet things...

The dimensionality curse phenomenon states that in high dimensional spaces distances between nearest and farthest points from query become almost equal. Therefore, neighbor calculations cannot discriminate candidate points. Many indexing methods try to cope with the have been proposed, but, usually these end up behaving like sequential scan over database terms of accessed pages when queries k-Nearest Neighbors are examined. In this paper, we experiment state art multi-attribute investigate...

This paper addresses the problem of head pose estimation with three degrees freedom (pitch, yaw, roll) from a single image. Pose is formulated as high-dimensional to low-dimensional mixture linear regression problem. We propose method that maps HOG-based descriptors, extracted face bounding boxes, corresponding poses. To account for errors in observed bounding-box position, we learn parameters such HOG descriptor mapped onto union and an offset, latter optimally shifts box towards actual...

The use of online videos is a common practice today amongst education professionals. Interactive features in are constantly evolving and recent trend the integration interactive elements web content into educational videos. paper (a) provides roadmap for using open source tools internet resources to develop learning environment where video aggregated with elements, educator coming from web, (b) describes how these used capturing storing learner activity data, and, (c) presents findings...

Virtual geographic environments long simulated real-world scenarios in urban planning, monument preservation, city infrastructure management, education, and entertainment. Their web-based visualisation distribution made these widely accessible. However, many systems remain static, lacking real-time data integration multi-user collaboration, while virtual worlds designed for the Metaverse emphasise dynamic interaction yet often omit essential geospatial context. Bridging this gap is critical...

Any multi-party conversation system benefits from speaker diarization, that is, the assignment of speech signals among participants. We here cast diarization problem into a tracking formulation whereby active is detected and tracked over time. A probabilistic tracker exploits on-image (spatial) coincidence visual auditory observations infers single latent variable which represents identity speaker. Both are explained by recently proposed weighted-data mixture model, while several options for...

This paper addresses the problem of video alignment. We present efficient approaches that allow for spatiotemporal alignment two sequences. Unlike most related works, we consider independently moving cameras capture a 3D scene at different times. The novelty proposed method lies in adaptation and extension an information retrieval framework casts sequences as image database set query frames, respectively. builds on recently quad descriptor. In this context, define Vote Space (VS) by...

Although the k-Nearest Neighbor classifier is one of most widely-used classification methods, it suffers from high computational cost and storage requirements involves. These major drawbacks have constituted an active research field over last decades. This paper proposes effective data reduction algorithm that has low preprocessing reduces while maintaining accuracy at acceptable level. The proposed based on a fast pre-processing clustering procedure creates homogeneous clusters. centroids...