Browsing by Author "Berretti, Stefano"

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    CAD 3D Model Classification by Graph Neural Networks: A new Approach based on STEP Format
    (The Eurographics Association, 2022) Mandelli, Lorenzo; Berretti, Stefano; Cabiddu, Daniela; Schneider, Teseo; Allegra, Dario; Catalano, Chiara Eva; Cherchi, Gianmarco; Scateni, Riccardo
    In this paper, we introduce a new approach for retrieval and classification of 3D models that directly performs in the Computer- Aided Design (CAD) format without any conversion to other representations like point clouds or meshes, thus avoiding any loss of information. Among the various CAD formats, we consider the widely used STEP extension, which represents a standard for product manufacturing information. This particular format represents a 3D model as a set of primitive elements such as surfaces and vertices linked together. In our approach, we exploit the linked structure of STEP files to create a graph in which the nodes are the primitive elements and the arcs are the connections between them. We then use Graph Neural Networks (GNNs) to solve the problem of model classification. Finally, we created two datasets of 3D models in native CAD format, respectively, by collecting data from the Traceparts model library and from the Configurators software modeling company. We used these datasets to test and compare our approach with respect to state-of-the-art methods that consider other 3D formats. Our code is available at https://github.com/divanoLetto/3D_STEP_Classification
  • Thumbnail Image
    Item
    Depth-Based Face Recognition by Learning from 3D-LBP Images
    (The Eurographics Association, 2019) Neto, Joao Baptista Cardia; Marana, Aparecido Nilceu; Ferrari, Claudio; Berretti, Stefano; Bimbo, Alberto Del; Biasotti, Silvia and Lavoué, Guillaume and Veltkamp, Remco
    In this paper, we propose a hybrid framework for face recognition from depth images, which is both effective and efficient. It consists of two main stages: First, the 3DLBP operator is applied to the raw depth data of the face, and used to build the corresponding descriptor images (DIs). However, such operator quantizes relative depth differences over/under +-7 to the same bin, so as to generate a fixed dimensional descriptor. To account for this behavior, we also propose a modification of the traditional operator that encodes depth differences using a sigmoid function. Then, a not-so-deep (shallow) convolutional neural network (SCNN) has been designed that learns from the DIs. This architecture showed two main advantages over the direct application of deep-CNN (DCNN) to depth images of the face: On the one hand, the DIs are capable of enriching the raw depth data, emphasizing relevant traits of the face, while reducing their acquisition noise. This resulted decisive in improving the learning capability of the network; On the other, the DIs capture low-level features of the face, thus playing the role for the SCNN as the first layers do in a DCNN architecture. In this way, the SCNN we have designed has much less layers and can be trained more easily and faster. Extensive experiments on low- and high-resolution depth face datasets confirmed us the above advantages, showing results that are comparable or superior to the state-of-the-art, using by far less training data, time, and memory occupancy of the network.
  • Thumbnail Image
    Item
    SHREC 2020 Track: River Gravel Characterization
    (The Eurographics Association, 2020) Giachetti, Andrea; Biasotti, Silvia; Moscoso Thompson, Elia; Fraccarollo, Luigi; Nguyen, Quang; Nguyen, Hai-Dang; Tran, Minh-Triet; Arvanitis, Gerasimos; Romanelis, Ioannis; Fotis, Vlasis; Moustakas, Konstantinos; Tortorici, Claudio; Werghi, Naoufel; Berretti, Stefano; Schreck, Tobias and Theoharis, Theoharis and Pratikakis, Ioannis and Spagnuolo, Michela and Veltkamp, Remco C.
    The quantitative analysis of the distribution of the different types of sands, gravels and cobbles shaping river beds is a very important task performed by hydrologists to derive useful information on fluvial dynamics and related processes (e.g., hydraulic resistance, sediment transport and erosion, habitat suitability. As the methods currently employed in the practice to perform this evaluation are expensive and time-consuming, the development of fast and accurate methods able to provide a reasonable estimate of the gravel distribution based on images or 3D scanning data would be extremely useful to support hydrologists in their work. To evaluate the suitability of state-of-the-art geometry processing tool to estimate the distribution from digital surface data, we created, therefore, a dataset including real captures of riverbed mockups, designed a retrieval task on it and proposed them as a challenge of the 3D Shape Retrieval Contest (SHREC) 2020. In this paper, we discuss the results obtained by the methods proposed by the groups participating in the contest and baseline methods provided by the organizers. Retrieval methods have been compared using the precision-recall curves, nearest neighbor, first tier, second tier, normalized discounted cumulated gain and average dynamic recall. Results show the feasibility of gravels characterization from captured surfaces and issues in the discrimination of mixture of gravels of different size.
  • No Thumbnail Available
    Item
    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.