Galanakis, G.; Zabulis, X. & Argyros, A. “Nearest Neighbor-Based Data Denoising for Deep Metric Learning“ in Proceedings of the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, vol.2, pp.595-603, 2024, https://zenodo.org/records/10630698
The effectiveness of supervised deep metric learning relies on the availability of a correctly annotated dataset, i.e., a dataset where images are associated with correct class labels. The presence of incorrect labels in a dataset disorients the learning process. In this paper, we propose an approach to combat the presence of such label noise in datasets. Our approach operates online, during training and on the batch level. It examines the neighborhood of samples, considers which of them are noisy and eliminates them from the current training step. The neighborhood is established using features obtained from the entire dataset during previous training epochs and therefore is updated as the model learns better data representations. We evaluate our approach using multiple datasets and loss functions, and demonstrate that it performs better or comparably to the competition. At the same time, in contrast to the competition, it does not require knowledge of the noise contamination rate of the employed datasets.
Bartalesi V., De Martino C. & Lenzi E. Using Large Language Models to Create Narrative Events. PeerJ Computer Science. 2024; 10:e2242: https://doi.org/10.7717/peerj-cs.2242
Narratives play a crucial role in human communication, serving as a means to convey experiences, perspectives, and meanings across various domains. They are particularly significant in scientific communities, where narratives are often utilized to explain complex phenomena and share knowledge. This article explores the possibility of integrating large language models (LLMs) into a workflow that, exploiting the Semantic Web technologies, transforms raw textual data gathered by scientific communities into narratives. In particular, we focus on using LLMs to automatically create narrative events, maintaining the reliability of the generated texts. The study provides a conceptual definition of narrative events and evaluates the performance of different smaller LLMs compared to the requirements we identified. A key aspect of the experiment is the emphasis on maintaining the integrity of the original narratives in the LLM outputs, as experts often review texts produced by scientific communities to ensure their accuracy and reliability. We first perform an evaluation on a corpus of five narratives and then on a larger dataset comprising 124 narratives. LLaMA 2 is identified as the most suitable model for generating narrative events that closely align with the input texts, demonstrating its ability to generate high-quality narrative events. Prompt engineering techniques are then employed to enhance the performance of the selected model, leading to further improvements in the quality of the generated texts.
Evdaimon T., Katzourakis A., Partarakis N., Xhako A., Zabulis X. & Zidianakis E. Reviving Antiquity in the Digital Era: Digitization, Semantic Curation, and VR Exhibition of Contemporary Dresses. Computers. 2024; 13(3), 57. https://doi.org/10.3390/computers1303005
In this paper, we present a comprehensive methodology to support the multifaceted process involved in the digitization, curation, and virtual exhibition of cultural heritage artifacts. The proposed methodology is applied in the context of a unique collection of contemporary dresses inspired by antiquity. Leveraging advanced 3D technologies, including lidar scanning and photogrammetry, we meticulously captured and transformed physical garments into highly detailed digital models. The postprocessing phase refined these models, ensuring an accurate representation of the intricate details and nuances inherent in each dress. Our collaborative efforts extended to the dissemination of this digital cultural heritage, as we partnered with the national aggregator in Greece, SearchCulture, to facilitate widespread access. The aggregation process streamlined the integration of our digitized content into a centralized repository, fostering cultural preservation and accessibility. Furthermore, we harnessed the power of these 3D models to transcend traditional exhibition boundaries, crafting a virtual experience that transcends geographical constraints. This virtual exhibition not only enables online exploration but also invites participants to immerse themselves in a captivating virtual reality environment. The synthesis of cutting-edge digitization techniques, cultural aggregation, and immersive exhibition design not only contributes to the preservation of contemporary cultural artifacts but also redefines the ways in which audiences engage with and experience cultural heritage in the digital age.
Bartalesi V., Demeridou I., Fallahian P., Meghini C., Nikolaou N., Partarakis N., Pratelli N. & Zabulis X. Modelling and Simulation of Traditional Craft Actions. Applied Sciences. 2024; 14(17), 7750. https://doi.org/10.3390/app14177750
The problem of modelling and simulating traditional crafting actions is addressed, motivated by the goals of craft understanding, documentation, and training. First, the physical entities involved in crafting actions are identified, physically, and semantically characterised, including causing entities, conditions, properties, and objects, as well as the space and time in which they occur. Actions are semantically classified into a taxonomy of four classes according to their goals, which are shown to exhibit similarities in their operation principles and utilised tools. This classification is employed to simplify the create archetypal simulators, based on the Finite Element Method, by developing archetypal simulators for each class and specialising them in craft-specific actions. The approach is validated by specialising the proposed archetypes into indicative craft actions and predicting their results in simulation. The simulated actions are rendered in 3D to create visual demonstrations and can be integrated into game engines for training applications.
Partarakis N, Zabulis X. A Review of Immersive Technologies, Knowledge Representation, and AI for Human-Centered Digital Experiences. Electronics. 2024; 13(2):269. https://doi.org/10.3390/electronics13020269
The evolution of digital technologies has resulted in the emergence of diverse interaction technologies. In this paper, we conducted a review of seven domains under a human-centric approach user interface design, human-centered web-based information systems, semantic knowledge representation, X-reality applications, human motion and 3D digitization, serious games, and AI. In this review, we studied these domains concerning their impact on the way we interact with digital interfaces, process information, and engage in immersive experiences. As such, we highlighted the shifts in design paradigms, user-centered principles, and the rise of web-based information systems. The results of such shifts are materialized in modern immersive technologies, semantic knowledge representation, serious games, and the facilitation of artificial intelligence for interactions. Through this exploration, we aimed to assist our understanding of the challenges that lie ahead. The seamless integration of technologies, ethical considerations, accessibility, education for technological literacy, interoperability, user trust, environmental sustainability, and regulatory frameworks are becoming significant. These challenges present opportunities for the future to enrich human experiences while addressing societal needs. This paper lays the groundwork for thoughtful and innovative approaches to the challenges that will define the future of human–computer interaction and information technologies.
Koutlemanis, P., Zabulis, X., Stivaktakis, N., Partarakis, N., Zidianakis, E., & Demeridou, I. A Low-Cost, Close-Range Photogrammetric Surface Scanner. Frontiers in Imaging, vol. 3. 2024, 1341343. https://doi.org/10.3389/fimag.2024.1341343
To achieve micrometer resolution in reconstruction, accurate and photorealistic surface digitization, and retain low manufacturing cost, an image acquisition approach and a reconstruction method are proposed. The image acquisition approach uses the CNC to systematically move the camera and acquire images in a grid tessellation and at multiple distances from the target surface. A relatively large number of images is required to cover the scanned surface. The reconstruction method tracks keypoint features to robustify correspondence matching and uses far-range images to anchor the accumulation of errors across a large number of images utilized.
Partarakis N, Zabulis X. Applying Cognitive Load Theory to eLearning of Crafts. Multimodal Technologies and Interaction. 2024; 8(1):2. https://doi.org/10.3390/mti8010002
Craft education and training are important for preserving cultural heritage and fostering artisanal skills. However, the pedagogical challenges in this domain are numerous. This research paper presents a comprehensive framework for applying Cognitive Load Theory to enhance craft education and training via eLearning platforms. In this study, practical guidelines based on CLT principles are provided to optimize the instructional design and content delivery. These guidelines scaffold craft learning experiences within eLearning platforms and encompass strategies to manage cognitive load, promote active learning, and facilitate gradual transition. Subsequently, the paper details the implementation of these guidelines within a popular eLearning platform, Moodle, emphasizing its adaptability and utility for craft education. It discusses the customization of Moodle courses to align with the cognitive load management principles, providing a practical blueprint for educators and instructional designers. The research culminates in a case study, wherein the guidelines are applied to a craft eLearning course using Moodle.
Demeridou I., Koutlemanis P., Partarakis N., Stivaktakis, N. M., Zabulis X. & Zidianakis E. A Close-Range Photogrammetric Surface Scanner and its Evaluation. Preprints. 2023, 202307.1609. https://www.preprints.org/manuscript/202307.1609/v1
The effectiveness of supervised deep metric learning relies on the availability of a correctly annotated dataset, i.e., a dataset where images are associated with correct class labels. The presence of incorrect labels in a dataset disorients the learning process. In this paper, we propose an approach to combat the presence of such label noise in datasets. Our approach operates online, during training and on the batch level. It examines the neighborhood of samples, considers which of them are noisy and eliminates them from the current training step. The neighborhood is established using features obtained from the entire dataset during previous training epochs and therefore is updated as the model learns better data representations. We evaluate our approach using multiple datasets and loss functions, and demonstrate that it performs better or comparably to the competition. At the same time, in contrast to the competition, it does not require knowledge of the noise contamination rate of the employed datasets.
Stavroulakis, P.I.; Ganetsos, T.; Zabulis, X. Large Scale Optical Projection Tomography without the Use of Refractive-Index-Matching Liquid. Sensors 2023, 23, 9814.https://doi.org/10.3390/s23249814
The practical, rapid, and accurate optical 3D reconstruction of transparent objects with contemporary non-contact optical techniques, has been an open challenge in the field of optical metrology. The combination of refraction, reflection, and transmission in transparent objects makes it very hard to use common off-the-shelf 3D reconstruction solutions to accurately reconstruct transparent objects in three dimensions without completely coating the object with an opaque material. We demonstrate in this work that a specific class of transparent objects can indeed be reconstructed without the use of opaque spray coatings, via Optical Projection Tomography (OPT). Particularly, the 3D reconstruction of large thin-walled hollow transparent objects can be achieved via OPT, without the use of refractive-index-matching liquid, accurately enough for use in both cultural heritage and beverage packaging industry applications. We compare 3D reconstructions of our proposed OPT method to those achieved by an industrial-grade 3D scanner and report average shape differences of ±0.34 mm for ‘shelled’ hollow objects and ±0.92 mm for ‘non-shelled’ hollow objects. A disadvantage of using OPT, which was noticed on the thicker ‘non-shelled’ hollow objects, as opposed to the ‘shelled’ hollow objects, was that it induced partial filling of hollow areas and the deformation of embossed features.
This entry discusses the challenge of safeguarding crafts in Europe. Safeguarding is defined herein as the systematic process of understanding, representing, preserving, and valorizing crafts following the recommendations of UNESCO and the UN-World Tourism Organization. The abovementioned challenges are discussed through a multidisciplinary prism starting from the scientific challenges in the information and communication technologies sector and expanding the discussion to ethical, legal, and policy-making measures and recommendations to safeguard crafts as a form of tangible and intangible cultural heritage but also as a source of growth and impact for the communities that practice them. To this end, the role of education and training for craft preservation is discussed, considering that the declining number of practitioners and apprentices is considered today the main threat to their preservation.
B. E. Olivas-Padilla, A. Glushkova and S. Manitsaris, “Motion Capture Benchmark of Real Industrial Tasks and Traditional Crafts for Human Movement Analysis,” in IEEE Access, vol. 11, pp. 40075-40092, 2023, https://doi.org/10.1109/ACCESS.2023.3269581.
Human movement analysis is a key area of research in robotics, biomechanics, and data science. It encompasses tracking, posture estimation, and movement synthesis. While numerous methodologies have evolved over time, a systematic and quantitative evaluation of these approaches using verifiable ground truth data of three-dimensional human movement is still required to define the current state of the art. This paper presents seven datasets recorded using inertial-based motion capture. The datasets contain professional gestures carried out by industrial operators and skilled craftsmen performed in real conditions in-situ. The datasets were created with the intention of being used for research in human motion modeling, analysis, and generation. The protocols for data collection are described in detail, and a preliminary analysis of the collected data is provided as a benchmark. The Gesture Operational Model, a hybrid stochastic-biomechanical approach based on kinematic descriptors, is utilized to model the dynamics of the experts’ movements and create mathematical representations of their motion trajectories for analyzing and quantifying their body dexterity. The models allowed accurate generation of human professional poses and an intuitive description of how body joints cooperate and change over time through the performance of the task.
Zabulis, X.; Partarakis, N.; Demeridou, I.; Doulgeraki, P.; Zidianakis, E.; Argyros, A.; Theodoridou, M.; Marketakis, Y.; Meghini, C.; Bartalesi, V.; et al. A Roadmap for Craft Understanding, Education, Training, and Preservation. Heritage 2023, 6, 5305-5328. https://doi.org/10.3390/heritage6070280
A roadmap is proposed that defines a systematic approach for craft preservation and its evaluation. The proposed roadmap aims to deepen craft understanding so that blueprints of appropriate tools that support craft documentation, education, and training can be designed while achieving preservation through the stimulation and diversification of practitioner income. In addition to this roadmap, an evaluation strategy is proposed to validate the efficacy of the developed results and provide a benchmark for the efficacy of craft preservation approaches. The proposed contribution aims at the catalyzation of craft education and training with digital aids, widening access and engagement to crafts, economizing learning, increasing exercisability, and relaxing remoteness constraints in craft learning.
