Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
Immersive RockArt: When rock carvings meet photogrammetry and computer graphics
Full text linkRock Art Immersive is an interdisciplinary project for the 3D documentation, valorisation, communication, and tourist promotion of the UNESCO site of the Pitoti rock carvings in Val Camonica (Italy). Photogrammetry and Computer Graphics are
coupled to create reality-based interactive and communicative material to safeguard and valorize a heritage site
Stories from the Peaks: An Interactive Data Storytelling to Narrate Climate Change Impacts through a Pluralism of Voices
No full textExplaining the seriousness of climate change while encouraging audiences to take action to counteract it is especially challenging. Data about temperature and CO2 increases are often perceived as abstract and ungraspable, while extreme events like floods and droughts generate a sense of awe and helplessness. This paper presents the design process of an interactive data storytelling prototype about climate change and overtourism in Trentino. Rooted in the principles of data humanism and feminist epistemologies, this
study emphasizes the importance of combining scientific data with locality and including a plurality of voices. Through a meticulous data curation process involving public institutions and scientific experts, the storytelling was designed to integrate human and more-than-human perspectives, grounded in rigorous and credible data. An initial evaluation with eleven users shows that data-driven narratives can effectively convey the complex challenges climate
change imposes on mountain communities and ecosystems
Coupling V-SLAM and Semantic Segmentation for Cultural Heritage Documentation
Full text link3D digitization has become an essential tool in cultural heritage documentation, offering unprecedented opportunities for preservation, analysis, and dissemination. Beyond only capturing 3D spatial geometry, the semantic enrichment of 3D models is rapidly evolving offering a more efficient interpretation and usage of 3D data. Traditionally, 3D semantic enrichment has relied on point cloud-based segmentation. However, 3D point cloud-based segmentation approaches can struggle with the efficient identification of small-scale, geometric elements, or visually ambiguous classes, limiting their applicability in such contexts. This study leverages the rich contextual and textural information of 2D imagery to detect challenging semantic categories, such as fine architectural elements (e.g., individual stone blocks) and material decay (e.g., material detachment and material loss), using deep learning-based 2D semantic segmentation techniques. These detections are then projected into 3D space through a 2D-to-3D semantic segmentation framework that couples V-SLAM and 3D results with the 2D predictions. The framework is evaluated on data acquired using the fish-eye multi-camera mobile mapping system ATOM-ANT3D in two challenging case study environments. Achieved results demonstrate a reliable level of accu-racy given the inherent complexity of targeted classes, enhancing the interpretability of 3D models by providing meaningful and met-rically interpreted objects classifications in 3D models. (Demonstration video: https://youtu.be/GidxhNS7ECc
Virtual Reality in Reminiscence Therapy for Parkinson’s Disease (RETURN-VR): A Co-Design Approach
No full text
Exploiting Symbolic Heuristics for the Synthesis of Domain-Specific Temporal Planning Guidance Using Reinforcement Learning
No full textRecent work investigated the use of Reinforcement Learning (RL) for the synthesis of heuristic guidance to improve the performance of temporal planners when a domain is fixed and a set of training problems (not plans) is given. The idea is to extract a heuristic from the value function of a particular (possibly infinite-state) MDP constructed over the training problems.
In this paper, we propose an evolution of this learning and planning framework that focuses on exploiting the information provided by symbolic heuristics during both the RL and planning phases. First, we formalize different reward schemata for the synthesis and use symbolic heuristics to mitigate the problems caused by the truncation of episodes needed to deal with the potentially infinite MDP. Second, we propose learning a residual of an existing symbolic heuristic, which is a "correction" of the heuristic value, instead of eagerly learning the whole heuristic from scratch. Finally, we use the learned heuristic in combination with a symbolic heuristic using a multiple-queue planning approach to balance systematic search with imperfect learned information.
We experimentally compare all the approaches, highlighting their strengths and weaknesses and significantly advancing the state of the art for this planning and learning schema
Virtual Reality in Reminiscence Therapy for Parkinson’s Disease (RETURN-VR): A Co-Design Approach.
No full text2024 JOSA A Emerging Researcher Best Paper Prize: editorial
Full text linkJOSA A Editor-in-Chief Olga Korotkova, Deputy Editor Markus Testorf, and the members of the 2024 Emerging Researcher Best Paper Prize Committee announce the recipient of the 2024 prize for the best paper published by an emerging researcher in the Journal
Curiosity Driven Multi-agent Reinforcement Learning for 3D Game Testing
No full textRecently testing of games via autonomous agents has shown great promise in tackling challenges faced by the game industry, which mainly relied on either manual testing or record/replay. In particular Reinforcement Learning (RL) solutions have shown potential by learning directly from playing the game without the need for human intervention.In this paper, we present cMarlTest, an approach for testing 3D games through curiosity driven Multi-Agent Reinforcement Learning (MARL). cMarlTest deploys multiple agents that work collaboratively to achieve the testing objective. The use of multiple agents helps resolve issues faced by a single agent approach.We carried out experiments on different levels of a 3D game comparing the performance of cMarlTest with a single agent RL variant. Results are promising where, considering three different types of coverage criteria, cMarlTest achieved higher coverage. cMarlTest was also more efficient in terms of the time taken, with respect to the single agent based variant
On the Impact of Hate Speech Synthetic Data on Model Fairness
Full text linkAlthough attention has been devoted to the issue of online hate speech, some phenomena, such as ableism or ageism, are scarcely represented by existing datasets and case studies. This can lead to hate speech detection systems that do not perform well on underrepresented identity groups. Given the unprecedented capabilities of LLMs in producing high-quality data, we investigate the possibility of augmenting existing data with generative language models, reducing target imbalance. We experiment with augmenting 1,000 posts from the Measuring Hate Speech corpus, an English dataset annotated with target identity information, adding around 30,000 synthetic examples using both simple data augmentation methods and different types of generative models, comparing autoregressive and sequence-to-sequence approaches. We focus our evaluation on the performance of models on different identity groups, finding that performance can differ greatly for different targets and "simpler" data augmentation approaches can improve classification better than state-of-the-art language models