Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
Filling the Hive: A Reflective Toolkit for Community-led Rural Development
No full textDesigning socio-technical systems for rural areas requires empowering local actors to actively engage in a creative process that thoroughly considers the unique characteristics of these territories. Methods intended for urban contexts often fail to account for rural communities’ specific values, cultures, and needs. Furthermore, the focus is usually on the design of the digital part of the innovation, its requirements, and technological constraints, leaving out other essential enablers. To address this, we present a toolkit designed to boost reflection on the key ‘'ingredients'’ — social, economic, technological, political, and infrastructural — essential for addressing rural challenges through socio-
technical interventions. The toolkit offers tangible, user- friendly resources that encourage dialogue, self-reflection, and the creative envisioning of rural transformations. It includes inspirational cards as well as reflection questions to support communities in evaluating their needs, resources, and aspirations, fostering self-awareness, learning, and action
Searches for direct slepton production in the compressed-mass corridor in √s = 13 TeV pp collisions with the ATLAS detector
No full textThis paper presents searches for the direct pair production of charged light-flavour sleptons, each decaying into a stable neutralino and an associated Standard Model lepton. The analyses focus on the challenging “corridor” region, where the mass difference, ∆m, between the slepton ( or ) and the lightest neutralino () is less or similar to the mass of the W boson, m(W), with the aim to close a persistent gap in sensitivity to models with ∆m ≲ m(W). Events are required to contain a high-energy jet, significant missing transverse momentum, and two same-flavour opposite-sign leptons (e or μ). The analysis uses pp collision data at
= 13 TeV recorded by the ATLAS detector, corresponding to an integrated luminosity of 140 fb−1. Several kinematic selections are applied, including a set of boosted decision trees. These are each optimised for different ∆m to provide expected sensitivity for the first time across the full ∆m corridor. The results are generally consistent with the Standard Model, with the most significant deviations observed with a local significance of 2.0 σ in the selectron search, and 2.4 σ in the smuon search. While these deviations weaken the observed exclusion reach in some parts of the signal parameter space, the previously present sensitivity gap to this corridor is largely reduced. Constraints at the 95% confidence level are set on simplified models of selectron and smuon pair production, where selectrons (smuons) with masses up to 300 (350) GeV can be excluded for ∆m between 2 GeV and 100 GeV
From Ship to Shore: Food, Drink, and Cultural Encounter in Gregorio Mengarini’s Flathead Nineteenth-Century Mission
Get PDFThis study examines the experiences of the Jesuit missionary Gregorio Mengarini among the Flathead people in the mid-nineteenth century, focusing on the practices of eating and the reasons for dietary changes in several different contexts. Drawing on Mengarini’s vocation and the shared cultural significance of food, the essay examines historical sources documenting his journey. The travels of Mengarini and his companions, both from Europe to the United States and within the US, highlight the challenges of hunger and dietary restrictions. The Jesuit mission with the Flathead people aimed to establish subsistence agriculture, which often clashed with the local food culture rooted in practices such as root gathering and buffalo hunting. These cultural differences presented considerable difficulties for the Catholic missionaries in adapting. Mengarini’s narrative also highlights the arduous conditions the Jesuits faced in the Rocky Mountains, including hunger and travel-related dangers. The article concludes with the eventual demise of the Flathead Mission, emphasizing the challenges associated with introducing a new way of life to indigenous communities. This study offers a nuanced perspective on cross-cultural interactions, subsistence strategies, and the limitations of missionary efforts in transforming traditional ways of life
3D Forest Semantic Segmentation Using Multispectral LiDAR and 3D Deep Learning
No full textRegular forest inventory is essential for conservation and management. Over the past decades, laser scanning has emerged as a remote and non-destructive solution to streamline this laborious process. Advanced multispectral (MS) laser scanning systems simultaneously acquire 3D spatial and spectral information across multiple wavelengths, enabling estimation of forest biophysical and biochemical traits. This study investigates the potential of airborne MS laser scanning for fine-grained forest semantic segmentation into six components (ground, low vegetation, trunk, branches, foliage, and woody debris), thereby supporting forest inventory and analysis. We evaluate three state-of-the-art 3D deep learning models (kernel point convolution (KPConv), superpoint transformer (SPT), and point transformer V3 (PTv3)) and random forest model. Our analysis reveals the superiority of PTv3, outperforming the other models by 21.8 percentage points (pp) with the mean intersection over union (mIoU) of 69.1%. Additionally, our rigorous spectral ablation study demonstrates that MS laser scanning data substantially improves the segmentation results, increasing the IoU of woody debris, branches, and trunks by 12.7 pp, 4.5 pp, and 2.5 pp, respectively. This study highlights the strong potential of MS laser scanning to enable automated and accurate forest inventory through prior fine-grained forest semantic segmentation
Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding
No full textA multi-sensor multi-resolution dataset to support forest inventory methods
Get PDFAccurate estimation of forest structural and taxonomic parameters is vital for biodiversity monitoring, carbon accounting and sustainable management. Most of the current methods for estimating these parameters are still developed and tested on site-specific case studies, limiting reproducibility and cross-site generalization. This paper introduces 3D3, a multi-sensor and multi-resolution benchmark dataset designed to evaluate 3D forestry algorithms across diverse European forest types. 3D3 includes data collected by airborne, helicopter, UAV and terrestrial (static and mobile) laser scanning systems along with RGB and hyperspectral imagery, covering a variety of forest types (Boreal, Alpine and Mediterranean). By encompassing both mono- and multi-wavelength laser data, 3D3 represents a unique resource for developing new algorithms and evaluating them on distinct datasets. Each site provides ground truth for at least one task among Individual Tree Segmentation (ITS), Forest Semantic Segmentation (FSS) or tree parameter estimation and species classification
Benchmarking Vectorized Building Footprint Extraction from Very High Resolution Aerial Imagery
Get PDFAccurate, topologically consistent building footprints are essential for building reconstruction and GIS applications. But highresolution orthophotos often contain occlusions (trees, cast shadows, etc.) or dense roof structures that challenge pixel-based segmentation and polygonization. In recent years, Line Segment Detection (LSD) networks have gained popularity as they can directly extract vectorized building footprints. This study benchmarks three line-segment detection (LSD) networks - L-CNN, ULSD, and FClip - against a strong semantic segmentation network - DeepLabV3+ - for building footprint extraction from very high resolution orthophotos across multiple regions with varied built-up morphology. Our evaluation on the considered urban areas revealed that LSD approaches generally deliver cleaner boundaries and more reliable roof topology than segmentation methods, whose high pixel scores mask boundary breaks. These findings indicate that when polygonal fidelity and downstream GIS usability are priorities, LSD pipelines could be superior for vectorized building footprint extraction compared to segmentation methods
Deconver: A Deconvolutional Network for Medical Image Segmentation
No full text: While convolutional neural networks (CNNs) and vision transformers (ViTs) have advanced medical image segmentation, they face inherent limitations such as local receptive fields in CNNs and high computational complexity in ViTs. This paper introduces Deconver, a novel network that integrates traditional deconvolution techniques from image restoration as a core learnable component within a U -shaped architecture. Deconver replaces computationally expensive attention mechanisms with efficient nonnegative deconvolution (NDC) operations, enabling the restoration of high-frequency details while suppressing artifacts. Key innovations include a backpropagation-friendly NDC layer based on a provably monotonic update rule and a parameter-efficient design. Evaluated across five datasets (ISLES'22, Spleen, BraTS'23, GlaS, and FIVES) covering both 2D and 3D segmentation tasks, Deconver achieves state-of-the-art performance in Dice scores and Hausdorff distance while reducing computational costs (FLOPs) by up to 90% compared to leading baselines. By bridging traditional image restoration with deep learning, this work offers a practical solution for high-precision segmentation in resource-constrained clinical workflows