2019
Journal article
Open Access
Towards a topological-geometrical theory of group equivariant non-expansive operators for data analysis and machine learning
Bergomi M. G., Frosini P., Giorgi D., Quercioli N.
We provide a general mathematical framework for group and set equivariance in machine learning. We define group equivariant non-expansive operators (GENEOs) as maps between function spaces associated with groups of transformations. We study the topological and metric properties of the space of GENEOs to evaluate their approximating power and set the basis for general strategies to initialize and compose operators. We define suitable pseudo-metrics for the function spaces, the equivariance groups and the set of non-expansive operators. We prove that, under suitable assumptions, the space of GENEOs is compact and convex. These results provide fundamental guarantees in a machine learning perspective. By considering isometry-equivariant non-expansive operators, we describe a simple strategy to select and sample operators. Thereafter, we show how selected and sampled operators can be used both to perform classical metric learning and to inject knowledge in artificial neural networks.Source: Nature Machine Intelligence 1 (2019): 423–433. doi:10.1038/s42256-019-0087-3
DOI: 10.1038/s42256-019-0087-3
DOI: 10.48550/arxiv.1812.11832
Project(s): 5HTCircuits
Metrics:
Bergomi M. G., Frosini P., Giorgi D., Quercioli N.
We provide a general mathematical framework for group and set equivariance in machine learning. We define group equivariant non-expansive operators (GENEOs) as maps between function spaces associated with groups of transformations. We study the topological and metric properties of the space of GENEOs to evaluate their approximating power and set the basis for general strategies to initialize and compose operators. We define suitable pseudo-metrics for the function spaces, the equivariance groups and the set of non-expansive operators. We prove that, under suitable assumptions, the space of GENEOs is compact and convex. These results provide fundamental guarantees in a machine learning perspective. By considering isometry-equivariant non-expansive operators, we describe a simple strategy to select and sample operators. Thereafter, we show how selected and sampled operators can be used both to perform classical metric learning and to inject knowledge in artificial neural networks.Source: Nature Machine Intelligence 1 (2019): 423–433. doi:10.1038/s42256-019-0087-3
DOI: 10.1038/s42256-019-0087-3
DOI: 10.48550/arxiv.1812.11832
Project(s): 5HTCircuits
Metrics:
See at:
arXiv.org e-Print Archive 
| Nature Machine Intelligence | Nature Machine Intelligence | Archivio istituzionale della ricerca - Alma Mater Studiorum Università di Bologna | ISTI Repository | doi.org 
| www.nature.com | CNR ExploRA
2013
Journal article
Restricted
Information-Theoretic Selection of High-Dimensional Spectral Features for Structural Recognition
Boyan B., Escolano F., Giorgi D., Biasotti S.
Pattern recognition methods often deal with samples consisting of thousands of features. Therefore, the reduction of their dimensionality becomes crucial to make the data sets tractable. Feature selection techniques remove the irrelevant and noisy features and select a subset of features which describe better the samples and produce a better classification performance. In this paper, we propose a novel feature selection method for supervised classification within an information-theoretic framework. Mutual information is exploited for measuring the statistical relation between a subset of features and the class labels of the samples. Traditionally it has been measured for ranking single features; however, in most data sets the features are not independent and their combination provides much more information about the class than the sum of their individual prediction power. We analyze the use of different estimation methods which bypass the density estimation and estimate entropy and mutual information directly from the set of samples. These methods allow us to efficiently evaluate multivariate sets of thousands of features. Within this framework we experiment with spectral graph features extracted from 3D shapes. Most of the existing graph classification techniques rely on the graph attributes. We use unattributed graphs to show what is the contribution of each spectral feature to graph classification. Apart from succeeding to classify graphs from shapes relying only on their structure, we test to what extent the set of selected spectral features are robust to perturbations of the dataset.Source: Computer vision and image understanding (Print) 117 (2013): 214–228. doi:10.1016/j.cviu.2012.11.007
DOI: 10.1016/j.cviu.2012.11.007
Metrics:
Boyan B., Escolano F., Giorgi D., Biasotti S.
Pattern recognition methods often deal with samples consisting of thousands of features. Therefore, the reduction of their dimensionality becomes crucial to make the data sets tractable. Feature selection techniques remove the irrelevant and noisy features and select a subset of features which describe better the samples and produce a better classification performance. In this paper, we propose a novel feature selection method for supervised classification within an information-theoretic framework. Mutual information is exploited for measuring the statistical relation between a subset of features and the class labels of the samples. Traditionally it has been measured for ranking single features; however, in most data sets the features are not independent and their combination provides much more information about the class than the sum of their individual prediction power. We analyze the use of different estimation methods which bypass the density estimation and estimate entropy and mutual information directly from the set of samples. These methods allow us to efficiently evaluate multivariate sets of thousands of features. Within this framework we experiment with spectral graph features extracted from 3D shapes. Most of the existing graph classification techniques rely on the graph attributes. We use unattributed graphs to show what is the contribution of each spectral feature to graph classification. Apart from succeeding to classify graphs from shapes relying only on their structure, we test to what extent the set of selected spectral features are robust to perturbations of the dataset.Source: Computer vision and image understanding (Print) 117 (2013): 214–228. doi:10.1016/j.cviu.2012.11.007
DOI: 10.1016/j.cviu.2012.11.007
Metrics:
See at:
Computer Vision and Image Understanding | www.sciencedirect.com | CNR ExploRA
2014
Report
Unknown
SEMEOTICONS - In-depth analysis of state-of-the-art for facial expression analysis
Colantonio S., Chiarugi F., Giannakakis G., Pediaditis M., Iatraki G., Christinaki E., Manousos D., Pampouchidou A., Marias K., Tsiknakis M., Giorgi D., Matuszewski B. J.
Human face can reveal precious information about the healthy or unhealthy status of individuals. SEMEOTICONS aims to exploit the human face as an indicator of individual's health status and translate the semeiotic code of the face into measurements and descriptors automatically evaluated by computerized applications. WP5, in specific, focuses on developing methods for the assessment of stress, anxiety and fatigue based on automatic facial expression analysis using video. Consequently, a complete and thorough analysis of self-contained approaches, specialized methods and reported research results for facial expression analysis, targeted at the detection of signs that disclose such psycho-physical states is performed.Source: SEMEOTICONS - Deliverable D5.1, 2014., 2014
Project(s): SEMEOTICONS
Colantonio S., Chiarugi F., Giannakakis G., Pediaditis M., Iatraki G., Christinaki E., Manousos D., Pampouchidou A., Marias K., Tsiknakis M., Giorgi D., Matuszewski B. J.
Human face can reveal precious information about the healthy or unhealthy status of individuals. SEMEOTICONS aims to exploit the human face as an indicator of individual's health status and translate the semeiotic code of the face into measurements and descriptors automatically evaluated by computerized applications. WP5, in specific, focuses on developing methods for the assessment of stress, anxiety and fatigue based on automatic facial expression analysis using video. Consequently, a complete and thorough analysis of self-contained approaches, specialized methods and reported research results for facial expression analysis, targeted at the detection of signs that disclose such psycho-physical states is performed.Source: SEMEOTICONS - Deliverable D5.1, 2014., 2014
Project(s): SEMEOTICONS
See at: CNR ExploRA
2021
Journal article
Open Access
Smart Working at CNR-ISTI in the COVID-19 Era
Scopigno R., Giorgi D.
The CNR Institute for Information Science and Technologies describes its experience in adapting to smart working, which has dramatically changed the institute's modus operandi for most of the year 2020.Source: ERCIM news (2021): 6–7.
Scopigno R., Giorgi D.
The CNR Institute for Information Science and Technologies describes its experience in adapting to smart working, which has dramatically changed the institute's modus operandi for most of the year 2020.Source: ERCIM news (2021): 6–7.
See at:
ercim-news.ercim.eu | ISTI Repository | CNR ExploRA
2012
Conference article
Restricted
The Hitchhiker's guide to the galaxy of mathematical tools for shape analysis
S. Biasotti, B. Falcidieno, D. Giorgi, M. Spagnuolo
A practical guide for researchers who are exploring the new frontiers of 3D shape analysis and managing the complex mathematical tools that most methods rely on. Many research solutions come from advances in pure and applied mathematics, as well as from re-reading classical mathematical theories. Managing these math tools is critical to understanding and solving current problems in 3D shape analysis. This course is designed to help mathematicians and scientists communicate in a world where boundaries between disciplines are (fortunately) blurred, so they can quickly find the right mathematical tools for a bright intuitive idea and strike a balance between theoretical rigor and computationally feasible solutions. The course presents basic concepts in differential geometry and proceeds to advanced topics in algebraic topology, always keeping an eye on their computational counterparts. It includes examples of applications to shape correspondence, symmetry detection, and shape retrieval that show how these mathematical concepts can be translated into practical solutions.Source: SIGGRAPH'12 - ACM SIGGRAPH 2012 Courses, Los Angeles, CA, USA, 5-9 July 2012
DOI: 10.1145/2343483.2343499
Metrics:
S. Biasotti, B. Falcidieno, D. Giorgi, M. Spagnuolo
A practical guide for researchers who are exploring the new frontiers of 3D shape analysis and managing the complex mathematical tools that most methods rely on. Many research solutions come from advances in pure and applied mathematics, as well as from re-reading classical mathematical theories. Managing these math tools is critical to understanding and solving current problems in 3D shape analysis. This course is designed to help mathematicians and scientists communicate in a world where boundaries between disciplines are (fortunately) blurred, so they can quickly find the right mathematical tools for a bright intuitive idea and strike a balance between theoretical rigor and computationally feasible solutions. The course presents basic concepts in differential geometry and proceeds to advanced topics in algebraic topology, always keeping an eye on their computational counterparts. It includes examples of applications to shape correspondence, symmetry detection, and shape retrieval that show how these mathematical concepts can be translated into practical solutions.Source: SIGGRAPH'12 - ACM SIGGRAPH 2012 Courses, Los Angeles, CA, USA, 5-9 July 2012
DOI: 10.1145/2343483.2343499
Metrics:
See at:
dl.acm.org | doi.org | CNR ExploRA
2013
Journal article
Restricted
PHOG: Photometric and Geometric Functions for Textured Shape Retrieval
Biasotti S., Cerri A., Giorgi D., Spagnuolo M.
In this paper we target the problem of textured 3D object retrieval. As a first contribution, we show how to include photometric information in the persistence homology setting, also proposing a novel theoretical result about multidimensional persistence spaces. As a second contribution, we introduce a generalization of the integral geodesic distance which fuses shape and color properties. Finally, we adopt a purely geometric description based on the selection of geometric functions that are mutually independent. The photometric, hybrid and geometric descriptions are combined into a signature, whose performance is tested on a benchmark dataset.Source: Computer graphics forum (Print) 32 (2013): 13–22. doi:10.1111/cgf.12168
DOI: 10.1111/cgf.12168
Metrics:
Biasotti S., Cerri A., Giorgi D., Spagnuolo M.
In this paper we target the problem of textured 3D object retrieval. As a first contribution, we show how to include photometric information in the persistence homology setting, also proposing a novel theoretical result about multidimensional persistence spaces. As a second contribution, we introduce a generalization of the integral geodesic distance which fuses shape and color properties. Finally, we adopt a purely geometric description based on the selection of geometric functions that are mutually independent. The photometric, hybrid and geometric descriptions are combined into a signature, whose performance is tested on a benchmark dataset.Source: Computer graphics forum (Print) 32 (2013): 13–22. doi:10.1111/cgf.12168
DOI: 10.1111/cgf.12168
Metrics:
See at:
Computer Graphics Forum | onlinelibrary.wiley.com | CNR ExploRA
2014
Conference article
Restricted
SHREC'14 Track: Retrieval and Classification on Textured 3D Models
Biasotti S., Cerri A., Abdelrahman M., Aono M., Ben Hamza A., El-Melegy M., Farag A., Garro V., Giachetti D., Giorgi D., Godil A., Li C., Liu Y., Martono H., Sanada C., Tatsuma A., Velasco-Forero S., Xu C.
This paper reports the results of the SHREC'14 track: Retrieval and classification on textured 3D models, whose goal is to evaluate the performance of retrieval algorithms when models vary either by geometric shape or texture, or both. The collection to search in is made of 572 textured mesh models, having a two-level classification based on geometry and texture. Together with the dataset, a training set of 96 models was provided. The track saw eight participants and the submission of 22 runs, to either the retrieval or the classification contest, or both. The evaluation results show a promising scenario about textured 3D retrieval methods, and reveal interesting insights in dealing with texture information in the CIELab rather than in the RGB colour space.Source: Eurographics Workshop on 3D Object Retrieval (2014), pp. 111–120, Strasbourg, France, April 6, 2014
DOI: 10.2312/3dor.20141057
Metrics:
Biasotti S., Cerri A., Abdelrahman M., Aono M., Ben Hamza A., El-Melegy M., Farag A., Garro V., Giachetti D., Giorgi D., Godil A., Li C., Liu Y., Martono H., Sanada C., Tatsuma A., Velasco-Forero S., Xu C.
This paper reports the results of the SHREC'14 track: Retrieval and classification on textured 3D models, whose goal is to evaluate the performance of retrieval algorithms when models vary either by geometric shape or texture, or both. The collection to search in is made of 572 textured mesh models, having a two-level classification based on geometry and texture. Together with the dataset, a training set of 96 models was provided. The track saw eight participants and the submission of 22 runs, to either the retrieval or the classification contest, or both. The evaluation results show a promising scenario about textured 3D retrieval methods, and reveal interesting insights in dealing with texture information in the CIELab rather than in the RGB colour space.Source: Eurographics Workshop on 3D Object Retrieval (2014), pp. 111–120, Strasbourg, France, April 6, 2014
DOI: 10.2312/3dor.20141057
Metrics:
See at: CNR ExploRA
2014
Book
Restricted
Mathematical tools for shape analysis and description
Biasotti S., Falcidieno B., Giorgi D., Spagnuolo M.
This book is a guide for researchers and practitioners to the new frontiers of 3D shape analysis and the complex mathematical tools most methods rely on. The target reader includes students, researchers and professionals with an undergraduate mathematics background, who wish to understand the mathematics behind shape analysis. The authors begin with a quick review of basic concepts in geometry, topology, differential geometry, and proceed to advanced notions of algebraic topology, always keeping an eye on the application of the theory, through examples of shape analysis methods such as 3D segmentation, correspondence, and retrieval. A number of research solutions in the field come from advances in pure and applied mathematics, as well as from the re-reading of classical theories and their adaptation to the discrete setting. In a world where disciplines (fortunately) have blurred boundaries, the authors believe that this guide will help to bridge the distance between theory and practice.DOI: 10.2200/s00588ed1v01y201407cgr016
Metrics:
Biasotti S., Falcidieno B., Giorgi D., Spagnuolo M.
This book is a guide for researchers and practitioners to the new frontiers of 3D shape analysis and the complex mathematical tools most methods rely on. The target reader includes students, researchers and professionals with an undergraduate mathematics background, who wish to understand the mathematics behind shape analysis. The authors begin with a quick review of basic concepts in geometry, topology, differential geometry, and proceed to advanced notions of algebraic topology, always keeping an eye on the application of the theory, through examples of shape analysis methods such as 3D segmentation, correspondence, and retrieval. A number of research solutions in the field come from advances in pure and applied mathematics, as well as from the re-reading of classical theories and their adaptation to the discrete setting. In a world where disciplines (fortunately) have blurred boundaries, the authors believe that this guide will help to bridge the distance between theory and practice.DOI: 10.2200/s00588ed1v01y201407cgr016
Metrics:
See at:
www.morganclaypool.com | CNR ExploRA
2016
Journal article
Open Access
Wize Mirror - a smart, multisensory cardio-metabolic risk monitoring system
Andreu Y., Chiarugi F., Colantonio S., Giannakakis G., Giorgi D., Henriquez P., Kazantzaki E., Manousos D., Kostas M., Matuszewski B. J., Pascali M. A., Pediaditis M., Raccichini G., Tsiknakis M.
In the recent years personal health monitoring systems have been gaining popularity, both as a result of the pull from the general population, keen to improve well-being and early detection of possibly serious health conditions and the push from the industry eager to translate the current significant progress in computer vision and machine learning into commercial products. One of such systems is the Wize Mirror, built as a result of the FP7 funded SEMEOTICONS (SEMEiotic Oriented Technology for Individuals CardiOmetabolic risk self-assessmeNt and Self-monitoring) project. The project aims to translate the semeiotic code of the human face into computational descriptors and measures, automatically extracted from videos, multispectral images, and 3D scans of the face. The multisensory platform, being developed as the result of that project, in the form of a smart mirror, looks for signs related to cardio-metabolic risks. The goal is to enable users to self-monitor their well-being status over time and improve their life-style via tailored user guidance. This paper is focused on the description of the part of that system, utilising computer vision and machine learning techniques to perform 3D morphological analysis of the face and recognition of psycho-somatic status both linked with cardio-metabolic risks. The paper describes the concepts, methods and the developed implementations as well as reports on the results obtained on both real and synthetic datasets.Source: Computer vision and image understanding (Print) 148 (2016): 3–22. doi:10.1016/j.cviu.2016.03.018
DOI: 10.1016/j.cviu.2016.03.018
Project(s): SEMEOTICONS
Metrics:
Andreu Y., Chiarugi F., Colantonio S., Giannakakis G., Giorgi D., Henriquez P., Kazantzaki E., Manousos D., Kostas M., Matuszewski B. J., Pascali M. A., Pediaditis M., Raccichini G., Tsiknakis M.
In the recent years personal health monitoring systems have been gaining popularity, both as a result of the pull from the general population, keen to improve well-being and early detection of possibly serious health conditions and the push from the industry eager to translate the current significant progress in computer vision and machine learning into commercial products. One of such systems is the Wize Mirror, built as a result of the FP7 funded SEMEOTICONS (SEMEiotic Oriented Technology for Individuals CardiOmetabolic risk self-assessmeNt and Self-monitoring) project. The project aims to translate the semeiotic code of the human face into computational descriptors and measures, automatically extracted from videos, multispectral images, and 3D scans of the face. The multisensory platform, being developed as the result of that project, in the form of a smart mirror, looks for signs related to cardio-metabolic risks. The goal is to enable users to self-monitor their well-being status over time and improve their life-style via tailored user guidance. This paper is focused on the description of the part of that system, utilising computer vision and machine learning techniques to perform 3D morphological analysis of the face and recognition of psycho-somatic status both linked with cardio-metabolic risks. The paper describes the concepts, methods and the developed implementations as well as reports on the results obtained on both real and synthetic datasets.Source: Computer vision and image understanding (Print) 148 (2016): 3–22. doi:10.1016/j.cviu.2016.03.018
DOI: 10.1016/j.cviu.2016.03.018
Project(s): SEMEOTICONS
Metrics:
See at:
Central Lancashire Online Knowledge | Computer Vision and Image Understanding | ISTI Repository | www.sciencedirect.com | CNR ExploRA
2016
Conference article
Restricted
3D objects exploration: guidelines for future research
Biasotti S., Falcidieno B., Giorgi D., Spagnuolo M.
Search engines provide the interface to interact with 3D object repositories, which are rapidly growing in both number and size. This position paper presents the current state of the art on 3D dataset navigation and 3D model retrieval. We discuss a number of challenges we consider as the main points to be tackled for developing effective 3D object exploration systems.Source: Eurographics Workshop on 3D Object Retrieval, pp. 9–12, Lisbona, Portugal, 8 May 2016
DOI: 10.2312/3dor.20161081
Metrics:
Biasotti S., Falcidieno B., Giorgi D., Spagnuolo M.
Search engines provide the interface to interact with 3D object repositories, which are rapidly growing in both number and size. This position paper presents the current state of the art on 3D dataset navigation and 3D model retrieval. We discuss a number of challenges we consider as the main points to be tackled for developing effective 3D object exploration systems.Source: Eurographics Workshop on 3D Object Retrieval, pp. 9–12, Lisbona, Portugal, 8 May 2016
DOI: 10.2312/3dor.20161081
Metrics:
See at:
diglib.eg.org | CNR ExploRA
2017
Journal article
Open Access
Persistent homology to analyse 3D faces and assess body weight gain
Giorgi D., Pascali M. A., Henriquez P., Matuszewski B. J., Colantonio S., Salvetti O.
In this paper, we analyse patterns in face shape variation due to weight gain. We propose the use of persistent homology descriptors to get geometric and topological information about the configuration of anthropometric 3D face landmarks. In this way, evaluating face changes boils down to comparing the descriptors computed on 3D face scans taken at different times. By applying dimensionality reduction techniques to the dissimilarity matrix of descriptors, we get a space in which each face is a point and face shape variations are encoded as trajectories in that space. Our results show that persistent homology is able to identify features which are well related to overweight and may help assessing individual weight trends. The research was carried out in the context of the European project SEMEOTICONS, which developed a multisensory platform which detects and monitors over time facial signs of cardio-metabolic risk.Source: The visual computer 33 (2017): 549–563. doi:10.1007/s00371-016-1344-7
DOI: 10.1007/s00371-016-1344-7
Project(s): SEMEOTICONS
Metrics:
Giorgi D., Pascali M. A., Henriquez P., Matuszewski B. J., Colantonio S., Salvetti O.
In this paper, we analyse patterns in face shape variation due to weight gain. We propose the use of persistent homology descriptors to get geometric and topological information about the configuration of anthropometric 3D face landmarks. In this way, evaluating face changes boils down to comparing the descriptors computed on 3D face scans taken at different times. By applying dimensionality reduction techniques to the dissimilarity matrix of descriptors, we get a space in which each face is a point and face shape variations are encoded as trajectories in that space. Our results show that persistent homology is able to identify features which are well related to overweight and may help assessing individual weight trends. The research was carried out in the context of the European project SEMEOTICONS, which developed a multisensory platform which detects and monitors over time facial signs of cardio-metabolic risk.Source: The visual computer 33 (2017): 549–563. doi:10.1007/s00371-016-1344-7
DOI: 10.1007/s00371-016-1344-7
Project(s): SEMEOTICONS
Metrics:
See at:
Central Lancashire Online Knowledge | ISTI Repository | The Visual Computer | link.springer.com | CNR ExploRA
2016
Report
Unknown
SEMEOTICONS - Algorithms and methods for facial expression analysis and psycho-physical status evaluation - Final release
Chiarugi F., Christinaki E., Giannakakis G., Kazantzaki E., Manousos D., Pediaditis M., Tsiknakis M., Danielis A., Pascali M. A., Giorgi D., Raccichini G.
Stress, anxiety and fatigue have impact both on the physical and mental health. Stress and anxiety for example are implicated in the onset or progression of immunological, cardiovascular, circulatory or neurodegenerative diseases. Therefore, they have been included ab initio into the list of factors that need to be assessed for a successful health monitoring in the context of SEMEOTICONS. This deliverable presents the outcome of the work of Tasks 5.3 and 5.4, which includes the research, development and implementation of non-intrusive measurement methods of features describing stress, anxiety and fatigue. These methods take advantage of the information of colour and motion of the human face as captured by a video camera. The extracted features describe eye lid activity and opening, eye gaze and eye bags, dark circles, head motion, mouth activity and yawning, the validity of which has been studied thoroughly.Source: Project report, SEMEOTICONS, Deliverable D5.3.2, 2016
Project(s): SEMEOTICONS
Chiarugi F., Christinaki E., Giannakakis G., Kazantzaki E., Manousos D., Pediaditis M., Tsiknakis M., Danielis A., Pascali M. A., Giorgi D., Raccichini G.
Stress, anxiety and fatigue have impact both on the physical and mental health. Stress and anxiety for example are implicated in the onset or progression of immunological, cardiovascular, circulatory or neurodegenerative diseases. Therefore, they have been included ab initio into the list of factors that need to be assessed for a successful health monitoring in the context of SEMEOTICONS. This deliverable presents the outcome of the work of Tasks 5.3 and 5.4, which includes the research, development and implementation of non-intrusive measurement methods of features describing stress, anxiety and fatigue. These methods take advantage of the information of colour and motion of the human face as captured by a video camera. The extracted features describe eye lid activity and opening, eye gaze and eye bags, dark circles, head motion, mouth activity and yawning, the validity of which has been studied thoroughly.Source: Project report, SEMEOTICONS, Deliverable D5.3.2, 2016
Project(s): SEMEOTICONS
See at: CNR ExploRA
2015
Report
Unknown
SEMEOTICONS - Algorithms and methods for facial expression analysis and psycho-physical status evaluation
Chiarugi F., Christinaki E., Giannakakis G., Iatraki G., Kazantzaki E., Manousos D., Marias K., Pampouchidou A., Pediaditis M., Tsiknakis M., Giorgi D., Pascali M. A., Raccichini G., Henriquez P. Matuszewski B. J.
This document reports on the progress achieved in Task 5.3 "Stress and Anxiety assessment" and T5.4 "Fatigue assessment". This report provides the description of the facial signs/expressions potentially related to the three psychosomatic states under investigation in WP5 and, specifically, anxiety, fatigue and stress. Furthermore, it describes the implemented first version of the algorithms for the extraction of these facial signs/expressions and the datasets used for the training and the testing, and provides an evaluation of their performances.Source: Project report, SEMEOTICONS, Deliverable D5.3.1, 2015
Project(s): SEMEOTICONS
Chiarugi F., Christinaki E., Giannakakis G., Iatraki G., Kazantzaki E., Manousos D., Marias K., Pampouchidou A., Pediaditis M., Tsiknakis M., Giorgi D., Pascali M. A., Raccichini G., Henriquez P. Matuszewski B. J.
This document reports on the progress achieved in Task 5.3 "Stress and Anxiety assessment" and T5.4 "Fatigue assessment". This report provides the description of the facial signs/expressions potentially related to the three psychosomatic states under investigation in WP5 and, specifically, anxiety, fatigue and stress. Furthermore, it describes the implemented first version of the algorithms for the extraction of these facial signs/expressions and the datasets used for the training and the testing, and provides an evaluation of their performances.Source: Project report, SEMEOTICONS, Deliverable D5.3.1, 2015
Project(s): SEMEOTICONS
See at: CNR ExploRA
2015
Report
Unknown
SEMEOTICONS - D4.3.1 Methods for bio-morphometric and colorimetric face characterisation
Giorgi D., Raccichini G., Pascali M. A.
The present document is the first deliverable of Task 4.3 - Bio-morphometric and colorimetric face characterization. The objective of Task 4.3 is to develop a set of software tools to perform bio-morphometric facial measurements based on 3D facial data acquired via a 3D scanner. The measurements are expected to detect and monitor over time facial changes correlated with cardio-metabolic risk. Colour information is expected to complement the characterization.Source: Project report, SEMEOTICONS, Deliverable D4.3.1, 2015
Project(s): SEMEOTICONS
Giorgi D., Raccichini G., Pascali M. A.
The present document is the first deliverable of Task 4.3 - Bio-morphometric and colorimetric face characterization. The objective of Task 4.3 is to develop a set of software tools to perform bio-morphometric facial measurements based on 3D facial data acquired via a 3D scanner. The measurements are expected to detect and monitor over time facial changes correlated with cardio-metabolic risk. Colour information is expected to complement the characterization.Source: Project report, SEMEOTICONS, Deliverable D4.3.1, 2015
Project(s): SEMEOTICONS
See at: CNR ExploRA
2017
Journal article
Open Access
Lip segmentation based on Lambertian shadings and morphological operators for hyper-spectral images
Danielis A., Giorgi D., Larsson M., Stromberg T., Colantonio S., Salvetti O.
Lip segmentation is a non-trivial task because the colour difference between the lip and the skin regions maybe not so noticeable sometimes. We propose an automatic lip segmentation technique for hyper-spectral images from an imaging prototype with medical applications. Contrarily to many other existing lip segmentation methods, we do not use colour space transformations to localise the lip area. As input image, we use for the first time a parametric blood concentration map computed by using narrow spectral bands. Our method mainly consists of three phases: (i) for each subject generate a subset of face images enhanced by different simulated Lambertian illuminations, then (ii) perform lip segmentation on each enhanced image by using constrained morphological operations, and finally (iii) extract features from Fourier-based modeled lip boundaries for selecting the lip candidate. Experiments for testing our approach are performed under controlled conditions on volunteers and on a public hyper-spectral dataset. Results show the effectiveness of the algorithm against low spectral range, moustache, and noise.Source: Pattern recognition 63 (2017): 355–370. doi:10.1016/j.patcog.2016.10.007
DOI: 10.1016/j.patcog.2016.10.007
Project(s): SEMEOTICONS
Metrics:
Danielis A., Giorgi D., Larsson M., Stromberg T., Colantonio S., Salvetti O.
Lip segmentation is a non-trivial task because the colour difference between the lip and the skin regions maybe not so noticeable sometimes. We propose an automatic lip segmentation technique for hyper-spectral images from an imaging prototype with medical applications. Contrarily to many other existing lip segmentation methods, we do not use colour space transformations to localise the lip area. As input image, we use for the first time a parametric blood concentration map computed by using narrow spectral bands. Our method mainly consists of three phases: (i) for each subject generate a subset of face images enhanced by different simulated Lambertian illuminations, then (ii) perform lip segmentation on each enhanced image by using constrained morphological operations, and finally (iii) extract features from Fourier-based modeled lip boundaries for selecting the lip candidate. Experiments for testing our approach are performed under controlled conditions on volunteers and on a public hyper-spectral dataset. Results show the effectiveness of the algorithm against low spectral range, moustache, and noise.Source: Pattern recognition 63 (2017): 355–370. doi:10.1016/j.patcog.2016.10.007
DOI: 10.1016/j.patcog.2016.10.007
Project(s): SEMEOTICONS
Metrics:
See at:
ISTI Repository | Pattern Recognition | www.sciencedirect.com | CNR ExploRA
2018
Report
Unknown
Metamolds: Computational design of silicone molds
Alderighi T., Malomo L., Giorgi D., Pietroni N., Bickel B., Cignoni P.
We propose a new method for fabricating digital objects through reusable silicone molds. Molds are generated by casting liquid silicone into custom 3D printed containers called metamolds. Metamolds automatically define the cuts that are needed to extract the cast object from the silicone mold. The shape of metamolds is designed through a novel segmentation technique, which takes into account both geometric and topological constraints involved in the process of mold casting. Our technique is simple, does not require to change the shape or topology of the input objects, and only requires off-the-shelf materials and technologies. We successfully tested our method on a set of challenging examples with complex shapes and rich geometric detail.Source: ISTI Technical reports, 2018
Project(s): EMOTIVE
Alderighi T., Malomo L., Giorgi D., Pietroni N., Bickel B., Cignoni P.
We propose a new method for fabricating digital objects through reusable silicone molds. Molds are generated by casting liquid silicone into custom 3D printed containers called metamolds. Metamolds automatically define the cuts that are needed to extract the cast object from the silicone mold. The shape of metamolds is designed through a novel segmentation technique, which takes into account both geometric and topological constraints involved in the process of mold casting. Our technique is simple, does not require to change the shape or topology of the input objects, and only requires off-the-shelf materials and technologies. We successfully tested our method on a set of challenging examples with complex shapes and rich geometric detail.Source: ISTI Technical reports, 2018
Project(s): EMOTIVE
See at: CNR ExploRA
2019
Report
Unknown
A survey on 3D shape segmentation with focus on digital fabrication
Filoscia I., Alderighi T. Cignoni P., Giorgi D., Malomo L.
Segmenting 3D objects into parts is fundamental to a number of applications in computer graphics, including parametrization, texture mapping, shape matching, morphing, multi-resolution modeling, mesh editing, compression and animation [22]. Broadly speaking, shape segmentation techniques can be divided into geometry-based and semantics-based techniques. Geometry-based segmentations aim to partition the object into parts which have well-defined geometric properties such as size, curvature, or distance to a fitting primitive like a plane. Semantics-based segmentations, in turn, aim at identifying parts which are either visually relevant or meaningful in a given context, such as functional parts on mechanical objects or body parts on human models. Recently, 3D segmentation also drawn attention as a tool for efficient fabrication. The decomposition of objects into parts, indeed, helps solving different issues related to fabrication, such as height field constraints, volume constraints and need for supporting structures. In this work we present a complete survey of segmentation techniques, also highlighting their strengths and weaknesses. Our aim is to produce a handy overview to people who want to approach the problem of segmentation, especially if they want to apply it to digital fabrication.Source: ISTI Technical reports, 2019
Filoscia I., Alderighi T. Cignoni P., Giorgi D., Malomo L.
Segmenting 3D objects into parts is fundamental to a number of applications in computer graphics, including parametrization, texture mapping, shape matching, morphing, multi-resolution modeling, mesh editing, compression and animation [22]. Broadly speaking, shape segmentation techniques can be divided into geometry-based and semantics-based techniques. Geometry-based segmentations aim to partition the object into parts which have well-defined geometric properties such as size, curvature, or distance to a fitting primitive like a plane. Semantics-based segmentations, in turn, aim at identifying parts which are either visually relevant or meaningful in a given context, such as functional parts on mechanical objects or body parts on human models. Recently, 3D segmentation also drawn attention as a tool for efficient fabrication. The decomposition of objects into parts, indeed, helps solving different issues related to fabrication, such as height field constraints, volume constraints and need for supporting structures. In this work we present a complete survey of segmentation techniques, also highlighting their strengths and weaknesses. Our aim is to produce a handy overview to people who want to approach the problem of segmentation, especially if they want to apply it to digital fabrication.Source: ISTI Technical reports, 2019
See at: CNR ExploRA
2019
Report
Unknown
A novel segmentation algorithm for support-free 3D printing
Filoscia I., Alderighi T., Cignoni P., Giorgi D., Malomo L.
Digital fabrication, and 3D printing in particular, are growing important in a variety of fields, from industry to medicine, from cultural heritage to art, as often they are more rapid and cheaper than traditional manufacturing techniques. In this context, our aim is to make it easier for people to print high-quality objects at home, even of complex shape, by incorporating into software some of the professional skills that are needed to fully exploit the potential of 3D printing A major limitation of FDM printers is that the material must be supported when it is deposited: bridge-like structures or hanging features, which are not supported by other object parts, often need additional support structures. Indeed, most printers can produce overhangs, but only up to a certain tolerance angle, usually in-between 30 and 60 degrees. To solve this problem, additional columns of material are built to support the parts in overhang. These supports need to be removed in a postprocessing step, which may cause imperfections on the surface, or even break thin parts. A possible solution, which we adopt in this work, is to segment the object into smaller parts which can be printed individually with no or minimal need for supports. The main drawback is that the decomposition introduces cuts on the object surface, in correspondence of the boundaries between parts. Such cuts can be as visually disturbing as the imperfections due to the removal of supports, or even more. Therefore, given a 3D mesh representing the input object, our aim is to develop a segmentation technique to partition the mesh into a small number of simpler parts, each of which can be printed with no or minimal support, and such that the boundaries between the parts (i.e., where the cuts in the object surface will be) affect the appearance of the printed model as little as possible. We pose the segmentation problem as a multi-labeling problem solved via functional minimization. In our formulation, the data points will be mesh elements (either faces or clusters of faces), and the labels will be potential printing directions. We will define an objective function that takes into account the area of supported regions and support footings, as well as the visual impact of the cuts, in terms of both their length and location on the surface. We formulate this multi-labeling problem as an Integer Linear Program (ILP), which can be solved using standard optimization packages such as Gurobi.Source: ISTI Technical reports, 2019
Filoscia I., Alderighi T., Cignoni P., Giorgi D., Malomo L.
Digital fabrication, and 3D printing in particular, are growing important in a variety of fields, from industry to medicine, from cultural heritage to art, as often they are more rapid and cheaper than traditional manufacturing techniques. In this context, our aim is to make it easier for people to print high-quality objects at home, even of complex shape, by incorporating into software some of the professional skills that are needed to fully exploit the potential of 3D printing A major limitation of FDM printers is that the material must be supported when it is deposited: bridge-like structures or hanging features, which are not supported by other object parts, often need additional support structures. Indeed, most printers can produce overhangs, but only up to a certain tolerance angle, usually in-between 30 and 60 degrees. To solve this problem, additional columns of material are built to support the parts in overhang. These supports need to be removed in a postprocessing step, which may cause imperfections on the surface, or even break thin parts. A possible solution, which we adopt in this work, is to segment the object into smaller parts which can be printed individually with no or minimal need for supports. The main drawback is that the decomposition introduces cuts on the object surface, in correspondence of the boundaries between parts. Such cuts can be as visually disturbing as the imperfections due to the removal of supports, or even more. Therefore, given a 3D mesh representing the input object, our aim is to develop a segmentation technique to partition the mesh into a small number of simpler parts, each of which can be printed with no or minimal support, and such that the boundaries between the parts (i.e., where the cuts in the object surface will be) affect the appearance of the printed model as little as possible. We pose the segmentation problem as a multi-labeling problem solved via functional minimization. In our formulation, the data points will be mesh elements (either faces or clusters of faces), and the labels will be potential printing directions. We will define an objective function that takes into account the area of supported regions and support footings, as well as the visual impact of the cuts, in terms of both their length and location on the surface. We formulate this multi-labeling problem as an Integer Linear Program (ILP), which can be solved using standard optimization packages such as Gurobi.Source: ISTI Technical reports, 2019
See at: CNR ExploRA
2015
Conference article
Open Access
Smart mirror where I stand, who is the leanest in the sand?
Saba M., Scateni R., Sorrentino F., Spano L. D., Colantonio S., Giorgi D., Magrini M., Salvetti O., Buonaccorsi N., Vitali I.
In this paper we introduce the Virtuoso project, which aims at creating a seamless interactive support for fitness and wellness activities in touristic resorts. The overall idea is to evaluate the current physical state of the user through a technology-enhanced mirror. We describe the state of the art technologies for building a smart mirror prototype. In addition, we compare different parameters for evaluating the user's physical state, considering the user's impact, the contact requirements and their cost. Finally we depict the planned setup and evaluation setting for the Virtuoso project.Source: 9th International Conference on Universal Access in Human-Computer Interaction, UAHCI 2015, pp. 364–373, Los Angeles, CA, USA, 2-7 August, 2015
DOI: 10.1007/978-3-319-20684-4_36
Metrics:
Saba M., Scateni R., Sorrentino F., Spano L. D., Colantonio S., Giorgi D., Magrini M., Salvetti O., Buonaccorsi N., Vitali I.
In this paper we introduce the Virtuoso project, which aims at creating a seamless interactive support for fitness and wellness activities in touristic resorts. The overall idea is to evaluate the current physical state of the user through a technology-enhanced mirror. We describe the state of the art technologies for building a smart mirror prototype. In addition, we compare different parameters for evaluating the user's physical state, considering the user's impact, the contact requirements and their cost. Finally we depict the planned setup and evaluation setting for the Virtuoso project.Source: 9th International Conference on Universal Access in Human-Computer Interaction, UAHCI 2015, pp. 364–373, Los Angeles, CA, USA, 2-7 August, 2015
DOI: 10.1007/978-3-319-20684-4_36
Metrics:
See at:
link.springer.com | ISTI Repository | doi.org | link.springer.com | CNR ExploRA
2020
Journal article
Closed Access
ReviewerNet: A visualization platform for the selection of academic reviewers
Salinas M., Giorgi D., Ponchio F., Cignoni P.
We propose ReviewerNet, an online, interactive visualization system aimed to improve the reviewer selection process in the academic domain. Given a paper submitted for publication, we assume that good candidate reviewers can be chosen among the authors of a small set of pertinent papers; ReviewerNet supports the construction of such set of papers, by visualizing and exploring a literature citation network. The system helps journal editors and Program Committee members to select reviewers that do not have any conflict-of-interest and are representative of different research groups, by visualising the careers and co-authorship relations of candidate reviewers. The system is publicly available, and is demonstrated in the field of Computer Graphics.Source: Computers & graphics 89 (2020): 77–87. doi:10.1016/j.cag.2020.04.006
DOI: 10.1016/j.cag.2020.04.006
Metrics:
Salinas M., Giorgi D., Ponchio F., Cignoni P.
We propose ReviewerNet, an online, interactive visualization system aimed to improve the reviewer selection process in the academic domain. Given a paper submitted for publication, we assume that good candidate reviewers can be chosen among the authors of a small set of pertinent papers; ReviewerNet supports the construction of such set of papers, by visualizing and exploring a literature citation network. The system helps journal editors and Program Committee members to select reviewers that do not have any conflict-of-interest and are representative of different research groups, by visualising the careers and co-authorship relations of candidate reviewers. The system is publicly available, and is demonstrated in the field of Computer Graphics.Source: Computers & graphics 89 (2020): 77–87. doi:10.1016/j.cag.2020.04.006
DOI: 10.1016/j.cag.2020.04.006
Metrics:
See at:
Computers & Graphics | www.sciencedirect.com | CNR ExploRA