Search

International audience; Person re-identification from videos taken by multiple cameras from different views is a very challenging problem that has attracted growing interest in last years. In fact, the same person from significant cross-view has different appearances from clothes change, illumination, and cluttered background. To deal with this issue, we use the skeleton information since it is not affected by appearance and pose variations. The skeleton as an input is projected on the Grassmann manifold in order to model the human motion as a trajectory. Then, we calculate the distance on the Grassmann manifold, in order to guarantee invariance against rotation, as well as local distances allowing to discriminate anthropometric for each person. The two distances are thereafter combined while defining dynamically the optimal combination weight for each person. Indeed, a machine learning process learns to predict the best weight for each person according to the rank metric of its re-identification results. Experimental results, using challenging 3D (IASLab RGBD-ID and BIWI-Lab RGBD-ID) and 2D (Prid-2011 and i-LIDS-VID) benchmarks, show that the proposed method can boost re-id ranking thanks to its ability to define the optimal weight for each person independently of view and pose changes.

La communauté IMT qui porte le colloque Risques et Cyber est en quelque sorte par nature transverse aux écoles des Mines et aux écoles des Télécom, en rassemblant des profils et des approches scientifiques assez différents. C’est donc une thématique qui illustre bien ce qu’on peut construire par le rassemblement des compétences de ces deux familles d'écoles. A travers l’ensemble des écoles, elle peut amener des compétences en cybersécurité, notamment en matière d’approches cybersécurité des systèmes, mais aussi en gestion de crise, que ce soit en simulation ou en modélisation, avec des approches numériques comme des approches relevant des sciences humaines et sociales. Si l’on ajoute les fortes compétences de nos écoles en ce qui concerne les systèmes industriels ou les systèmes de santé, mais aussi en modélisation et en optimisation, nous sommes capables d’apporter des réponses assez larges sur l’ensemble du spectre qui va de la détection à la résilience et à la remédiation, en passant par la gestion de crise.

In this paper, an effective pipeline to automatic 4D Facial Expression Recognition (4D FER) is proposed. It combines two growing but disparate ideas in Computer Vision -- computing the spatial facial deformations using tools from Riemannian geometry and magnifying them using temporal filtering. The flow of 3D faces is first analyzed to capture the spatial deformations based on the recently-developed Riemannian approach, where registration and comparison of neighboring 3D faces are led jointly. Then, the obtained temporal evolution of these deformations are fed into a magnification method in order to amplify the facial activities over the time. The latter, main contribution of this paper, allows revealing subtle (hidden) deformations which enhance the emotion classification performance. We evaluated our approach on BU-4DFE dataset, the state-of-art 94.18% average performance and an improvement that exceeds 10% in classification accuracy, after magnifying extracted geometric features (deformations), are achieved., Comment: International Conference On Pattern Recognition 2016

Human activity recognition is one of the most challenging and active areas of research in the computer vision domain. However, designing automatic systems that are robust to significant variability due to object combinations and the high complexity of human motions are more challenging. In this paper, we propose to model the inter-frame rigid evolution of skeleton parts as the trajectory in the Lie group SE(3)&times, &hellip, &times, SE(3). The motion of the object is similarly modeled as an additional trajectory in the same manifold. The classification is performed based on a rate-invariant comparison of the resulting trajectories mapped to a vector space, the Lie algebra. Experimental results on three action and activity datasets show that the proposed method outperforms various state-of-the-art human activity recognition approaches.

International audience; The security of persons requires a beefy guarantee in our society particularly, with the spread of terrorism nowadays throughout the world. In this context, palmprint identification based on texture analysis is amongst the triumphal pattern recognition applications to recognize the persons. In this paper, we investigated a deep texture analysis for the palmprint texture pattern representation based on a fusion between several texture information extracting through multiple descriptors, such as "HOG and Gabor Filters", "Fractal dimensions" and "GLCM" corresponding respectively to the Frequency, Model, and Statistical methodologies-based texture features. We assessed the proposed deep texture analysis method as well as the applicability of the dimensionality reduction techniques and the correlation concept between the features-based fusion on the challenging PolyU, CASIA and IIT-Delhi Palmprint databases. The experimental results show that the fusion of different texture types using the correlation concept for palmprint modality identification leads to promising results.

Human action recognition in 3D sequences is one of the most challenging and active areas of research in the computer vision domain. However designing automatic systems that are robust to significant variability due to object combinations and high complexity of human motions is more challenging in addition to the typical requirements such as rotation, translation, and scale invariance is challenging task. In this paper, we propose a spatio-temporal modeling of human-object interaction videos for online and off-line recognition. The inter-joint distances and the object are considered as low-level features for online classification. For off-line recognition, we propose rate-invariant classification of full video and early recognition. A shape analysis of trajectories of the inter-joint and object-joints distances is proposed for this end. The experiments conducted following state-of-the-art settings using MSR Daily Activity 3D Dataset and Online RGBD Action Dataset and on a new multi-view dataset for human object interaction demonstrate that the proposed approach is effective and discriminative for human object interaction classification as demonstrated here.

The detection and tracking of human landmarks in video streams has gained in reliability partly due to the availability of affordable RGB-D sensors. The analysis of such time-varying geometric data is playing an important role in the automatic human behavior understanding. However, suitable shape representations as well as their temporal evolution, termed trajectories, often lie to nonlinear manifolds. This puts an additional constraint (i.e., nonlinearity) in using conventional Machine Learning techniques. As a solution, this paper accommodates the well-known Sparse Coding and Dictionary Learning approach to study time-varying shapes on the Kendall shape spaces of 2D and 3D landmarks. We illustrate effective coding of 3D skeletal sequences for action recognition and 2D facial landmark sequences for macro- and micro-expression recognition. To overcome the inherent nonlinearity of the shape spaces, intrinsic and extrinsic solutions were explored. As main results, shape trajectories give rise to more discriminative time-series with suitable computational properties, including sparsity and vector space structure. Extensive experiments conducted on commonly-used datasets demonstrate the competitiveness of the proposed approaches with respect to state-of-the-art., Comment: 14 pages, 5 figures

We propose an efficient method for principal line extraction from the palmprint and geometric framework for analyzing their shapes. This representation, along with the elastic Riemannian metric, seems natural for measuring principal line deformations and is robust to challenges such as orientation variation and re-parameterization due to pose variation and missing part, respectively. The palmprint texture is investigated using the fractal analysis; thus the resulting features are fused with the principal line features. This framework is shown to be promising from both – empirical and theoretical – perspectives. In terms of empirical evaluation, our results match or improve the state-of-the-art methods on three prominent palmprint datasets: PolyU, CASIA, and IIT-Delhi, each posing a different type of challenge. From a theoretical perspective, this framework allows fusing texture analysis and shape analysis.

Despite a great deal of progress in face recognition technologies, current solutions are still vulnerable to spoof attacks. In fact, it is easy to access digital replicas of facial biometric information from readily available photos, videos and 3D masks. The literature contains several face anti spoofing methods that try to detect whether the face in the front of the recognition system is real or an artificial replica. However, these methods are not robust and require many improvements since they are sensitive to lightening conditions and pose variations. In order to address these issues, we propose a novel face anti spoofing method based on Multi Color Convolutional Neural Network (CNN) architecture named DeepColorFASD. Our approach investigates the effect of space colors (RGB, HSV and Y CbCr) on CNN architectures and proposes a fusion based voting method for face anti spoofing. In addition, we also explain the resulting feature maps visualizations. We evaluate our system through an experimental study using CASIA FASD: a well-known face anti spoofing database. The results using this challenging database demonstrate that our solution performs better than recent works as measured by Half Total Error Rate (HTER) and ROC curve.

International audience; Suitable shape representations as well as their temporal evolution, termed trajectories, often lie to non-linear mani-folds. This puts an additional constraint (i.e., non-linearity) in using conventional machine learning techniques for the purpose of classification, event detection, prediction, etc. This paper accommodates the well-known Sparse Coding and Dictionary Learning to the Kendall's shape space and illustrates effective coding of 3D skeletal sequences for action recognition. Grounding on the Riemannian geometry of the shape space, an intrinsic sparse coding and dictionary learning formulation is proposed for static skeletal shapes to overcome the inherent non-linearity of the manifold. As a main result, initial trajectories give rise to sparse code functions with suitable computational properties, including sparsity and vector space representation. To achieve action recognition, two different classification schemes were adopted. A bi-directional LSTM is directly performed on sparse code functions, while a linear SVM is applied after representing sparse code functions using Fourier temporal pyramid. Experiments conducted on three publicly available datasets show the superiority of the proposed approach compared to existing Riemannian representations and its competitiveness with respect to other recently-proposed approaches. When the benefits of invariance are maintained from the Kendall's shape representation, our approach not only overcomes the problem of non-linearity but also yields to discriminative sparse code functions.

This paper describes a novel intra-modal feature fusion for palmprint recognition based on fusing multiple descriptors to analyze the complex texture pattern. The main contribution lies in the combination of several texture features extracted by the Multi-descriptors, namely: Gabor Filters, Fractal Dimension and Gray Level Concurrence Matrix. This means to their effectiveness to confront the various challenges in terms of scales, position, direction and texture deformation of palmprint in unconstrained environments. The extracted Gabor filter-based texture features from the preprocessed palmprint images to be fused with the Fractal dimension-based-texture features and Gray Level Concurrence Matrix-based texture features using the Multiset Canonical Correlation Analysis method (MCCA). Realized experiments on three benchmark datasets prove that the proposed method surpasses other well-known state of the art methods and produces encouraging recognition rates by reaching 97.45% and 96.93% for the PolyU and IIT-Delhi Palmprint datasets.

International audience; In this paper, we are interested in people re-identification using skeleton information provided by a consumer RGB-D sensor. We perform the modelling and the analysis of human motion by focusing on 3D human joints given by skeletons. In fact, the motion dynamic is modeled by projecting skeleton information on Grassmann manifold. Moreover, in order to define the identity of a test trajectory, we compare it against a labeled trajectory database while using an unsupervised similarity assessment procedure. Indeed, the main contribution of this work resides in the introduced distance that combines temporal information as well as global and local geometrical ones. Realized experiments on standard datasets prove that the proposed method performs accurately even though it does not assume any prior knowledge.