Italian Chapter Conference 2023 - Smart Tools and Apps in Graphics

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/3543892

Browse

Now showing 1 - 18 of 18

Adjoint Bijective ZoomOut: Efficient Upsampling for Learned Linearly-invariant Embedding
(The Eurographics Association, 2023) Viganò, Giulio; Melzi, Simone; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
In this paper, we present a novel method for refining correspondences between 3D point clouds. Our method is compatible with the functional map framework, so it relies on the spectral representation of the correspondence. Although, differently from other similar approaches, this algorithm is specifically for a particular functional setting, being the only refinement method compatible with a recent data-driven approach, more suitable for point cloud matching. Our algorithm arises from a different way of converting functional operators into point-to-point correspondence, which we prove to promote bijectivity between maps, exploiting a theoretical result. Iterating this procedure and performing spectral upsampling in the same way as other similar methods, ours increases the accuracy of the correspondence, leading to more bijective correspondences. We tested our method over different datasets. It outperforms the previous methods in terms of map accuracy in all the tests considered.
An Approach to the Decomposition of Solids with Voids via Morse Theory
(The Eurographics Association, 2023) Pareja-Corcho, Juan; Montoya-Zapata, Diego; Moreno, Aitor; Cadavid, Carlos; Posada, Jorge; Arenas-Tobon, Ketzare; Ruiz-Salguero, Oscar; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
The decomposition of solids is a problem of interest in areas of engineering such as feature recognition or manufacturing planning. The problem can be stated as finding a set of smaller and simpler pieces that glued together amount to the initial solid. This decomposition can be guided by geometrical or topological criteria and be applied to either surfaces or solids (embedded manifolds). Most topological decompositions rely on Morse theory to identify changes in the topology of a manifold. A Morse function f is defined on the manifold and the manifold's topology is studied by studying the behaviour of the critical points of f . A popular structure used to encode this behaviour is the Reeb graph. Reeb graph-based decompositions have proven to work well for surfaces and for solids without inner voids, but fail to consider solids with inner voids. In this work we present a methodology based on the handle-decomposition of a manifold that can encode changes in the topology of solids both with and without inner voids. Our methodology uses the Boundary Representation of the solid and a shape similarity criteria to identify changes in the topology of both the outer and inner boundary(ies) of the solid. Our methodology is defined for Morse functions that produce parallel planar level sets and we do not consider the case of annidated solids (i.e. solids within other solids). We present an algorithm to implement our methodology and execute experiments on several datasets. Future work includes the testing of the methodology with functions different to the height function and the speed up of the algorithm's data structure.
AvatarizeMe: A Fast Software Tool for Transforming Selfies into Animatable Lifelike Avatars Using Machine Learning
(The Eurographics Association, 2023) Manfredi, Gilda; Capece, Nicola; Erra, Ugo; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Creating realistic avatars that faithfully replicate facial features from single-input images is a challenging task in computer graphics, virtual communication, and interactive entertainment. These avatars have the potential to revolutionize virtual experiences by enhancing user engagement and personalization. However, existing methods, such as 3D facial capture systems, are costly and complex. Our approach adopts the 3D Morphable Face Model (3DMM) method to create avatars with remarkably realistic features in a bunch of seconds, using only a single input image. Our method extends beyond facial shape resemblance; it meticulously generates both facial and bodily textures, enhancing overall likeness. Within Unreal Engine 5, our avatars come to life with real-time body and facial animations. This is made possible through a versatile skeleton for body and head movements and a suite of 52 face blendshapes, enabling the avatar to convey emotions and expressions with fidelity. This poster presents our approach, bridging the gap between reality and virtual representation, and opening doors to immersive virtual experiences with lifelike avatars.
Computational Design of Fabricable Geometric Patterns
(The Eurographics Association, 2023) Scandurra, Elena; Laccone, Francesco; Malomo, Luigi; Callieri, Marco; Cignoni, Paolo; Giorgi, Daniela; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
This paper addresses the design of surfaces as assemblies of geometric patterns with predictable performance in response to mechanical stimuli. We design a family of tileable and fabricable patterns represented as triangle meshes, which can be assembled for creating surface tessellations. First, a regular recursive subdivision of the planar space generates different geometric configurations for candidate patterns, having interesting and varied aesthetic properties. Then, a refinement step addresses manufacturability by solving for non-manifold configurations and sharp angles which would produce disconnected or fragile patterns. We simulate our patterns to evaluate their mechanical response when loaded in different scenarios targeting out-of-plane bending. Through a simple browsing interface, we show that our patterns span a variety of different bending behaviors. The result is a library of patterns with varied aesthetics and predefined mechanical behavior, to use for the direct design of mechanical metamaterials. To assess the feasibility of our approach, we show a pair of fabricated 3D objects with different curvatures.
A Foveated Framework to Accelerate Real-time Path Tracing
(The Eurographics Association, 2023) Mohanto, Bipul; Kluge, Sven; Staadt, Oliver; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
We developed a framework to accelerate real-time path tracing through foveated rendering, a robust technique that leverages human vision. Our dynamic foveated path-tracing framework integrates fixations and selectively lowers the rendering resolution towards the periphery. Through comprehensive experimentation, we demonstrated the effectiveness of our framework in this paper. Our solution can significantly enhance rendering performance, up to 25× without any notable visual differences. We further evaluated the framework using a structured error map algorithm with variable sample numbers.
A Gaze Detection System for Neuropsychiatric Disorders Remote Diagnosis Support
(The Eurographics Association, 2023) Cangelosi, Antonio; Antola, Gabriele; Iacono, Alberto Lo; Santamaria, Alfonso; Clerico, Marinella; Al-Thani, Dena; Agus, Marco; Calì, Corrado; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Accurate and early diagnosis of neuropsychiatric disorders, such as Autism Spectrum Disorders (ASD) is a significant challenge in clinical practice. This study explores the use of real-time gaze tracking as a tool for unbiased and quantitative analysis of eye gaze. The results of this study could support the diagnosis of disorders and potentially be used as a tool in the field of rehabilitation. The proposed setup consists of an RGB-D camera embedded in the latest-generation smartphones and a set of processing components for the analysis of recorded data related to patient interactivity. The proposed system is easy to use and doesn't require much knowledge or expertise. It also achieves a high level of accuracy. Because of this, it can be used remotely (telemedicine) to simplify diagnosis and rehabilitation processes. We present initial findings that show how real-time gaze tracking can be a valuable tool for doctors. It is a non-invasive device that provides unbiased quantitative data that can aid in early detection, monitoring, and treatment evaluation. This study's findings have significant implications for the advancement of ASD research. The innovative approach proposed in this study has the potential to enhance diagnostic accuracy and improve patient outcomes.
GPU-Accelerating Hierarchical Descriptors for Point Set Registration
(The Eurographics Association, 2023) Dutta, Somnath; Russig, Benjamin; Gumhold, Stefan; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
We present a GPU-accelerated global registration method for registering partial shapes, a common and often performancecritical task in many robotics, vision, and graphics applications. Global registration based on descriptor matching is highly dependent on the quality at which a shape is sampled, and computing expressive descriptors typically incurs high computation time. In this paper, we augment a global pair-wise registration algorithm based on hierarchical shape descriptors with a GPU-accelerated descriptor construction process, reducing the time spent on building descriptors by an order of magnitude. This allows for building more expressive descriptors, achieving a dual gain in both performance and accuracy. We conducted extensive evaluations on a large set of pair-wise registration problems, demonstrating very competitive registration accuracy, often rendering subsequent refinement with a local method unnecessary.
JPEG Line-drawing Restoration With Masks
(The Eurographics Association, 2023) Zhu, Yan; Yamaguchi, Yasushi; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Learning-based JPEG restoration methods usually lack consideration on the visual content of images. Even though these methods achieve satisfying results on photos, the direct application of them on line drawings, which consist of lines and white background, is not suitable. The large area of background in digital line drawings does not contain intensity information and should be constantly white (the maximum brightness). Existing JPEG restoration networks consistently fail to output constant white pixels for the background area. What's worse, training on the background can negatively impact the learning efficiency for areas where texture exists. To tackle these problems, we propose a line-drawing restoration framework that can be applied to existing state-of-the-art restoration networks. Our framework takes existing restoration networks as backbones and processes an input rasterized JPEG line drawing in two steps. First, a proposed mask-predicting network predicts a binary mask which indicates the location of lines and background in the potential undeteriorated line drawing. Then, the mask is concatenated with the input JPEG line drawing and fed into the backbone restoration network, where the conventional L1 loss is replaced by a masked Mean Square Error (MSE) loss. Besides learning-based mask generation, we also evaluate other direct mask generation methods. Experiments show that our framework with learnt binary masks achieves both better visual quality and better performance on quantitative metrics than the state-of-the-art methods in the task of JPEG line-drawing restoration.
Leveraging Moving Parameterization and Adaptive THB-Splines for CAD Surface Reconstruction of Aircraft Engine Components
(The Eurographics Association, 2023) Giannelli, Carlotta; Imperatore, Sofia; Mantzaflaris, Angelos; Mokriš, Dominik; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Reconstruction of highly accurate CAD models from point clouds is both paramount and challenging in industries such as aviation. Due to the acquisition process, this kind of data can be scattered and affected by noise, yet the reconstructed geometric models are required to be compact and smooth, while simultaneously capturing key geometric features of the engine parts. In this paper, we present an iterative moving parameterization approach, which consists of alternating steps of surface fitting, parameter correction, and adaptive refinement using truncated hierarchical B-splines (THB-splines). We revisit two existing surface fitting methods, a global least squares approximation and a hierarchical quasi-interpolation scheme, both based on THB-splines. At each step of the adaptive loop, we update the parameter locations by solving a non-linear optimization problem to infer footpoints of the point cloud on the current fitted surface. We compare the behavior of different optimization settings for the critical task of distance minimization, by also relating the effectiveness of the correction step to the quality of the initial parameterization. In addition, we apply the proposed approach in the reconstruction of aircraft engine components from scanned point data. It turns out that the use of moving parameterization instead of fixed parameter values, when suitably combined with the adaptive spline loop, can significantly improve the resulting surfaces, thus outperforming state-of-the-art hierarchical spline model reconstruction schemes.
Mixed Reality for Orthopedic Elbow Surgery Training and Operating Room Applications: A Preliminary Analysis
(The Eurographics Association, 2023) Cangelosi, Antonio; Riberi, Giacomo; Salvi, Massimo; Molinari, Filippo; Titolo, Paolo; Agus, Marco; Calì, Corrado; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
The use of Mixed Reality in medicine is widely documented to be a candidate to revolutionize surgical interventions. In this paper we present a system to simulate k-wire placement, that is a common orthopedic procedure used to stabilize fractures, dislocations, and other traumatic injuries.With the described system, it is possible to leverage Mixed Reality (MR) and advanced visualization techniques applied on a surgical simulation phantom to enhance surgical training and critical orthopedic surgical procedures. This analysis is centered on evaluating the precision and proficiency of k-wire placement in an elbow surgical phantom, designed with a 3D modeling software starting from a virtual 3D anatomical reference. By visually superimposing 3D reconstructions of internal structures and the target K-wire positioning on the physical model, it is expected not only to improve the learning curve but also to establish a foundation for potential real-time surgical guidance in challenging clinical scenarios. The performance is measured as the difference between K-wires real placement in respect to target position; the quantitative measurements are then used to compare the risk of iatrogenic injury to nerves and vascular structures of MRguided vs non MR-guided simulated interventions.
Remote and Deviceless Manipulation of Virtual Objects in Mixed Reality
(The Eurographics Association, 2023) Caputo, Ariel; Bartolomioli, Riccardo; Giachetti, Andrea; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Deviceless manipulation of virtual objects in mixed reality (MR) environments is technically achievable with the current generation of Head-Mounted Displays (HMDs), as they track finger movements and allow you to use gestures to control the transformation. However, when the object manipulation is performed at some distance, and when the transform includes scaling, it is not obvious how to remap the hand motions over the degrees of freedom of the object. Different solutions have been implemented in software toolkits, but there are still usability issues and a lack of clear guidelines for the interaction design. We present a user study evaluating three solutions for the remote translation, rotation, and scaling of virtual objects in the real environment without using handheld devices. We analyze their usability on the practical task of docking virtual cubes on a tangible shelf from varying distances. The outcomes of our study show that the usability of the methods is strongly affected by the use of separate or integrated control of the degrees of freedom, by the use of the hands in a symmetric or specialized way, by the visual feedback, and by the previous experience of the users.
A Scale-space Approach to the Morphological Simplification of Scalar Fields
(The Eurographics Association, 2023) Rocca, Luigi; Puppo, Enrico; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
We present a multi-scale morphological model of scalar fields based on the analysis of the spatial frequencies of the underlying function. Morphological models partition the domain of a function into homogeneous regions. The most popular tool in this field is the Morse-Smale complex, where each region is spanned by all integral lines that join a minimum to a maximum, with the integral lines departing from saddles as region boundaries. Morphological features usually occur at very different scales, from noise and high frequency details up to large trends at the lowest frequencies. Without some form of multi-scale analysis, only the morphology at the finest scale is visible and explicit in such a model. The most popular approach in the literature is the filtration provided by persistent homology, a method that combines the amplitude values of critical points with the topology of the sublevel sets of the function. We propose the adoption of an alternative filtration method, based on the analysis of the deep structure of the linear scale-space of the function. To retrieve an adequately fine-grained ranked sequence of pairs of critical points that vanish through the scales, we adopt a continuous representation of the scale-space that overcomes the limits of discrete scale-space approaches. This sequence provides a progressive simplification of the Morse-Smale complex, resulting in a progressive multi-scale model of the morphology that always refers to the geometry of the original function, which is not changed by our model. We apply our method to digital elevation models, with results providing a multi-scale representation of the network of ridges and valley lines that joins peaks, pits and passes and divide the land into mountains and basins.
Semantic Segmentation of High-resolution Point Clouds Representing Urban Contexts
(The Eurographics Association, 2023) Romanengo, Chiara; Cabiddu, Daniela; Pittaluga, Simone; Mortara, Michela; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
Point clouds are becoming an increasingly common digital representation of real-world objects, and they are particularly efficient when dealing with large-scale objects and/or when extremely high-resolution is required. The focus of our work is on the analysis, 3D feature extraction and semantic annotation of point clouds representing urban scenes, coming from various acquisition technologies, e.g., terrestrial (fixed or mobile) or aerial laser scanning or photogrammetry; the task is challenging, due to data dimensionality and noise. In particular, we present a pipeline to segment high-resolution point clouds representing urban environments into geometric primitives; we focus on planes, cylinders and spheres, which are the main features of buildings (walls, roofs, arches, ...) and ground surfaces (streets, pavements, platforms), and identify the unique parameters of each instance. This paper focuses on the semantic segmentation of buildings, but the approach is currently being generalised to manage extended urban areas. Given a dense point cloud representing a specific building, we firstly apply a binary space partitioning method to obtain small enough sub-clouds that can be processed. Then, a combination of the well-known RANSAC algorithm and a recognition method based on the Hough transform (HT) is applied to each sub-cloud to obtain a semantic segmentation into salient elements, like façades, walls and roofs. The parameters of primitive instances are saved as metadata to document the structural element of buildings for further thematic analyses, e.g., energy efficiency. We present a case study on the city of Catania, Italy, where two buildings of historical and artistic value have been digitized at very high resolution. Our approach is able to semantically segment these huge point clouds and it proves robust to uneven sampling density, input noise and outliers.
Smart Tools and Applications in Graphics - Eurographics Italian Chapter Conference: Frontmatter
(The Eurographics Association, 2023) Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
A Sparse Mesh Sampling Scheme for Graph-based Relief Pattern Classification
(The Eurographics Association, 2023) Paolini, Gabriele; Guiducci, Niccolò; Tortorici, Claudio; Berretti, Stefano; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
In the context of geometric deep learning, the classification of relief patterns involves recognizing the surface characteristics of a 3D object, regardless of its global shape. State-of-the-art methods leverage powerful 2D deep learning image-based techniques by converting local patches of the surface into a texture image. However, their effectiveness is guaranteed only when the mesh is simple enough to allow this projection onto a 2D subspace. Therefore, developing deep learning techniques that can work directly on manifolds represents an interesting line of research for addressing these challenges. The objective of our paper is to extend and enhance the architecture described in a recent GNN approach for a relief pattern classifier through the introduction of a new sampling tecnhique for meshes. In their method, local mesh structures, referred to as SpiderPatches, are connected to form the nodes of a graph, called MeshGraph, that captures global structures of the mesh. These two data structures are then fed into a bi-level architecture based on Graph Attention Networks. The MeshGraph construction proves important in ensuring optimal classification results. By the proposed subsampling process, we tackle the problem of fine-tuning multiple hyperparameters inherent the MeshGraph by defining a graph structure that is aware of the mesh geometric details. We demonstrate that the graph constructed using this approach robustly captures the relief patterns on the surface, obviating the need for data augmentation during training. The resulting network is robust, easily customizable, and shows comparable performance to recent methods, all while operating directly on 3D data.
Spectral-based Segmentation for Functional Shape-matching
(The Eurographics Association, 2023) Mancinelli, Claudio; Melzi, Simone; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
In Computer Graphics and Computer Vision, shape co-segmentation and shape-matching are fundamental tasks with diverse applications, from statistical shape analysis to human-robot interaction. These problems respectively target establishing segmentto- segment and point-to-point correspondences between shapes, which are crucial task for numerous practical scenarios. Notably, co-segmentation can aid in point-wise correspondence estimation in shape-matching pipelines like the functional maps framework. Our paper introduces an innovative shape segmentation pipeline which provides coherent segmentation for shapes within the same class. Through comprehensive evaluation on a diverse test set comprising shapes from various datasets and classes, we demonstrate the coherence of our segmentation approach. Moreover, our method significantly improves accuracy in shape matching scenarios, as evidenced by comparisons with the original functional maps approach. Importantly, these enhancements come with minimal computational overhead. Our work not only introduces a novel coherent segmentation method and a valuable tool for improving correspondence accuracy within functional maps, but also contributes to the theoretical foundations of this impactful field, inspiring further research.
User-assisted Sphere-mesh Construction
(The Eurographics Association, 2023) Paolillo, Davide; Taroni, Andrea; Tarini, Marco; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
In this study, we investigate the semi-automated generation of sphere-meshes as high-quality approximations for given threedimensional shapes, originally represented as common triangular meshes. A sphere-mesh is a class of geometric proxy defined as the volume swept by spheres with linearly interpolated centers and radii, that potentially strikes a good balance between conciseness of representation, simplicity of spatial queries, and expressive power, and is amenable to animations. Despite these favorable characteristics, its broader adoption in applications such as video games, physical simulation, or robotics is hindered by the difficulty of its construction, which remains an open problem. Existing fully automatic algorithms, based on interactive coarsening of the input mesh, fail to consistently produce satisfactory results, especially when very coarse sphere-meshes are sought. We improve on this situation with a 3D interface specifically designed to permit users to easily and intuitively modify the automatically generated models. The two phases (existing automatic algorithm and novel interactive tool), used in cascade, constitute a viable semi-automatic way to produce sphere-meshes. We test our method on several inputs tri-meshes, assess their quality, and finally experiment with a few downstream applications to exemplify the usability of our results.
VarIS: Variable Illumination Sphere for Facial Capture, Model Scanning, and Spatially Varying Appearance Acquisition
(The Eurographics Association, 2023) Baron, Jessica; Li, Xiang; Joshi, Parisha; Itty, Nathaniel; Greene, Sarah; Dhillon, Daljit Singh J.; Patterson, Eric; Banterle, Francesco; Caggianese, Giuseppe; Capece, Nicola; Erra, Ugo; Lupinetti, Katia; Manfredi, Gilda
We introduce VarIS, our Variable Illumination Sphere – a multi-purpose system for acquiring and processing real-world geometric and appearance data for computer-graphics research and production. Its key applications among many are (1) human-face capture, (2) model scanning, and (3) spatially varying material acquisition. Facial capture requires high-resolution cameras at multiple viewpoints, photometric capabilities, and a swift process due to human movement. Acquiring a digital version of a physical model is somewhat similar but with different constraints for image processing and more allowable time. Each requires detailed estimations of geometry and physically based shading properties. Measuring spatially varying light-scattering properties requires spanning four dimensions of illumination and viewpoint with angular, spatial, and spectral accuracy, and this process can also be assisted using multiple, simultaneous viewpoints or rapid switching of lights with no movement necessary. VarIS is a system of hardware and software for spherical illumination and imaging that has been custom designed and developed by our team. It has been inspired by Light Stages and goniophotometers, but costs less through use of primarily off-the-shelf components, and additionally extends capabilities beyond these devices. In this paper we describe the unique system and contributions, including practical details that could assist other researchers and practitioners.

Browse

Browsing Italian Chapter Conference 2023 - Smart Tools and Apps in Graphics by Title

Results Per Page

Sort Options