In recent years, various Automated Driving Systems (ADS) leveraging machine learning (ML) and deep learning (DL) have been developed to enhance passenger comfort and safety. These systems can monitor and record accident-prone events, such as the behavior of road users and abnormal occurrences, using data from sensors placed at different locations on the vehicle. However, similar advancements are still in the early stages within the railway domain, where automotive applications must be adapted for trains, requiring appropriate sensor systems and vehicle-specific adjustments. This article reviews the primary approaches to vision-based live risk assessment (VLRA) for light rail or very light rail systems vehicles that bridge the gap between cars and trains, representing an intermediate stage in transportation technology. The survey categorizes the main approaches based on the type of risk factor, learning-based methods used, evaluation metrics, specific sensors or processes involved, and relevant datasets. Finally, we provide conclusions and propose directions for future research.

Accurate egocentric body-pose estimation is essential for delivering immersive experiences in Virtual, Augmented, and Mixed Reality (VR/AR/MR) applications. A major challenge in this setting arises from the limited field of view provided by head-mounted cameras, which often leads to self-occlusion and poor visibility of certain body joints-particularly in the lower body. Current state-of-the-art methods typically perform pose estimation frame-by-frame in a view-aligned manner. While this is computationally efficient and performs well for visible joints, it struggles to estimate occluded joints due to the lack of contextual information. In this work, we explore the role of temporal information in improving joint estimation. We propose a lightweight model that refines pose predictions by leveraging a large temporal window. Our results show that incorporating such module improves the accuracy of joint estimation without incurring a large computation overhead.

For many years now, understanding the brain mechanism has been a great research subject in many different fields. Brain signal processing and especially electroencephalogram (EEG) has recently known a growing interest both in academia and industry. One of the main examples is the increasing number of Brain-Computer Interfaces (BCI) aiming to link brains and computers. In this paper, we present a novel framework allowing us to retrieve the attention state, i.e degree of attention given to a specific task, from EEG signals. While previous methods often consider the spatial relationship in EEG through electrodes and process them in recurrent or convolutional based architecture, we propose here to also exploit the time and frequency information with a transformer-based network that has already shown its supremacy in many machine-learning (ML) related studies, e.g. machine translation. In addition to this novel architecture, an extensive study on the feature extraction methods, frequential bands and temporal windows length has also been carried out. The proposed network has been trained and validated on two public datasets and achieves higher results compared to state-of-the-art models. As well as proposing better results, the framework could be used in real applications, e.g. Attention Deficit Hyperactivity Disorder (ADHD) symptoms or vigilance during a driving assessment.

Attention Deficit Hyperactivity Disorder (ADHD) is the most prevalent neurodevelopmental disorder among children. It affects patients’ lives in many ways: inattention, difficulty with stimuli inhibition or motor function regulation. Different treatments exist today, but these can present side effects or are not effective for all subgroups. Neurofeedback (NF) is an innovative treatment consisting of brain activity display. NF training could consist of a virtual reality (VR) video-game in which the participant’s attention affects the game. Attention being assessed through physiological signals, one of the main steps is to design an estimator for the attention state. We present a novel framework able to record physiological signals in specific attention states and able to estimate the corresponding attention state. We propose a database composed of electroencephalography signals (EEG), and an eye-tracker labelled with a score representing the attention span for 32 healthy participants. Different features are extracted from the signals and machine learning (ML) algorithms are proposed. Our approach exhibits high accuracy for attention estimation, which corroborates a correlation between attention state and physiological signals (i.e. EEG, eye-tracking signals). The dataset has been made publicly available to promote research in the domain and we encourage other scientists to use their own approach for attention estimation.

Eye-gaze information is used in a variety of user platforms, such as driver monitoring systems and head-mounted interfaces. In order to estimate human eye-gaze, many solutions have been proposed, using different devices and techniques. However, achieving such estimation using only cheap devices like RGB cameras would enable gaze interactions on mobile devices and therefore generalise this kind of interaction. It could also enable behavior studies based on gaze and made on every day devices. We propose in this paper a new method for eye-gaze estimation using a new deep learning architecture based on the Capsule Neural Network. Capsule Networks have shown great results so far on classification tasks, but only a few works use them for regression tasks.By taking advantage of the Capsule Network architecture and its ability to reconstruct images, we are able to recreate simplified eye images and then estimate human gaze from them. Experiments are performed on two representative datasets for the task of eye-gaze estimation. Encouraging results are obtained for both the estimation and the reconstruction.

The meaning of the wordstyledepends on its context. While actions have already been quite studied for a while,stylein human body motion is a growing topic of interest. In the context of animation,styleis crucial as it brings realism and expressiveness to the motion of a character. Even though it is undoubtedly a key element in motions, its definition and the use of the wordstylein itself, among research works, lack consensus. Achieving realistic motions is tedious. It requires either a large motion capture dataset or the considerable work of artist animators. The lack of consistentstyledata is thus a challenge. Stylistic motion generation is quite studied in order to overcome this issue. This paper focuses on the study ofstylein human body motion from 3D human body skeletal data. It establishes a taxonomy of definitions ofstyle, describes the data that have been used up until now, introduces key notions about motion capture data as well as machine learning, and presents approaches aboutstylerecognition, person identification through theirstyleand motionstylegeneration.

Action recognition in continuous video streams is a growing field since the past few years. Deep learning techniques and in particular Convolutional Neural Networks (CNNs) achieved good results in this topic. However, intrinsic CNNs limitations begin to cap the results since 2D CNN cannot capture temporal information and 3D CNN are to much resource demanding for real-time applications. Capsule Network, evolution of CNN, already proves its interesting benefits on small and low informational datasets like MNIST but yet its true potential has not emerged. In this paper we tackle the action recognition problem by proposing a new architecture combining Temporal Shift module over deep Capsule Network. Temporal Shift module permits us to insert temporal information over 2D Capsule Network with a zero computational cost to conserve the lightness of 2D capsules and their ability to connect spatial features. Our proposed approach outperforms or brings near state-of-the-art results on color and depth information on public datasets like First Person Hand Action and DHG 14/28 with a number of parameters 10 to 40 times less than existing approaches.

The need for efficient tools to index and retrieve 3D content becomes even more acute. This paper presents a fully automatic 3D-object retrieval method. It consists of two main steps namely shape signature extraction to describe the shape of objects, and similarity computing to compute similarity between objects. In the first step (signature extraction), we use a shape descriptor called geodesic cords. This descriptor can be seen as a probability distribution sampled from a shape function. In the second step (similarity computing), a global distance, based on belief function theory, is computed between each pair wise of descriptors corresponding respectively to an object query and an object from a given database. Experiments on commonly-used benchmarks demonstrate that our method obtains competitive performance compared to 3D-object retrieval methods from the state-of-the-art.

Grouping 3D objects into (semantically) meaningful categories is a challenging and important problem in 3D mining and shape processing. Here, we present a novel approach to categorize 3D objects. The method described in this article, is a belief-function-based approach and consists of two stages: the training stage, where 3D objects in the same category are processed and a set of representative parts is constructed, and the labeling stage, where unknown objects are categorized. The experimental results obtained on the Tosca-Sumner and the Shrec07 datasets show that the system efficiently performs in categorizing 3D models.

In this paper, we present a new method for 3D-shape categorization using Bag-of-Feature techniques (BoF). This method is based on vector quantization of invariant descriptors of 3D-object patches. We analyze the performance of two wellknown classifiers: the Naïve Bayes and the SVM. The results show the effectiveness of our approach and prove that the method is robust to non-rigid and deformable shapes, in which the class of transformations may be very wide due to the capability of such shapes to bend and assume different forms.

Abstract This paper presents a 3D‐mesh segmentation algorithm based on a learning approach. A large database of manually segmented 3D‐meshes is used to learn a boundary edge function. The function is learned using a classifier which automatically selects from a pool of geometric features the most relevant ones to detect candidate boundary edges. We propose a processing pipeline that produces smooth closed boundaries using this edge function. This pipeline successively selects a set of candidate boundary contours, closes them and optimizes them using a snake movement. Our algorithm was evaluated quantitatively using two different segmentation benchmarks and was shown to outperform most recent algorithms from the state‐of‐the‐art.

The 3D-shape matching problem plays a crucial role in many applications, such as indexing or modeling, by example. Here, we present a novel approach to matching 3D objects in the presence of nonrigid transformation and partially similar models. In this paper, we use the representation of surfaces by 3D curves extracted around feature points. Indeed, surfaces are represented with a collection of closed curves, and tools from shape analysis of curves are applied to analyze and to compare curves. The belief functions are used to define a global distance between 3D objects. The experimental results obtained on the TOSCA and the SHREC07 data sets show that the system performs efficiently in retrieving similar 3D models.

In this paper, we propose a new approach for 3D-shape matching. This approach encloses an off-line step and an on-line step. In the off-line one, an alphabet, of which any shape can be composed, is constructed. First, 3D-objects are subdivided into a set of 3D-parts. The subdivision consists to extract from each object a set of feature points with associated curves. Then the whole set of 3D-parts is clustered into different classes from a semantic point of view. After that, each class is modeled by a Hidden Markov Model (HMM). The HMM, which represents a character in the alphabet, is trained using the set of curves corresponding to the class parts. Hence, any 3D-object can be represented by a set of characters. The on-line step consists to compare the set of characters representing the 3D-object query and that of each object in the given dataset. The experimental results obtained on the TOSCA dataset show that the system efficiently performs in retrieving similar 3D-models.

In this paper we present a subjective quality assessment experiment for 3D-mesh segmentation. For this end, we carefully designed a protocol with respect to several factors namely the rendering conditions, the possible interactions, the rating range, and the number of human subjects. To carry out the subjective experiment, more than 40 human observers have rated a set of 250 segmentation results issued from various algorithms. The obtained Mean Opinion Scores, which represent the human subjects' point of view toward the quality of each segmentation, have then been used to evaluate both the quality of automatic segmentation algorithms and the quality of similarity metrics used in recent mesh segmentation benchmarking systems.

In this paper, we present an evaluation method of 3D-mesh segmentation algorithms based on a ground-truth corpus. This corpus is composed of a set of 3D-models grouped in different classes (animals, furnitures, etc.) associated with several manual segmentations produced by human observers. We define a measure that quantifies the consistency between two segmentations of a 3D-model, whatever their granularity. Finally, we propose an objective quality score for the automatic evaluation of 3D-mesh segmentation algorithms based on these measures and on the ground-truth corpus. Thus the quality of segmentations obtained by automatic algorithms is evaluated in a quantitative way thanks to the quality score, and on an objective basis thanks to the groundtruth corpus. Our approach is illustrated through the evaluation of two recent 3D-mesh segmentation methods.

Abstract This paper presents a novel approach for fast and efficient partial shape retrieval on a collection of 3D shapes. Each shape is represented by a Reeb graph associated with geometrical signatures. Partial similarity between two shapes is evaluated by computing a variant of their maximum common sub‐graph. By investigating Reeb graph theory, we take advantage of its intrinsic properties at two levels. First, we show that the segmentation of a shape by a Reeb graph provides charts with disk or annulus topology only. This topology control enables the computation of concise and efficient sub‐part geometrical signatures based on parameterisation techniques. Secondly, we introduce the notion of Reeb pattern on a Reeb graph along with its structural signature. We show this information discards Reeb graph structural distortion and still depicts the topology of the related sub‐parts. The number of combinations to evaluate in the matching process is then dramatically reduced by only considering the combinations of topology equivalent Reeb patterns. The proposed framework is invariant against rigid transformations and robust against non‐rigid transformations and surface noise. It queries the collection in interactive time (from 4 to 30 seconds for the largest queries). It outperforms the competing methods of the SHREC 2007 contest in term of NDCG vector and provides, respectively, a gain of 14.1% and 40.9% on the approaches by Biasotti et al.[ BMSF06 ]and Cornea et al.[ CDS*05 ]. As an application, we present an intelligent modelling‐by‐example system which enables a novice user to rapidly create new 3D shapes by composing shapes of a collection having similar sub‐parts.

Shape skeleton extraction is a fundamental pre-processing task in shape-based pattern recognition. This paper presents a new algorithm for fast and precise extraction of kinematic skeletons of 3D dynamic surface meshes. Unlike previous approaches, surface motions are characterized by the mesh edge-length deviation induced by its transformation through time. Then a static skeleton extraction algorithm based on Reeb graphs exploits this latter information to extract the kinematic skeleton. This hybrid static and dynamic shape analysis enables the precise detection of objects¿ articulations as well as shape topological transitions corresponding to possibly-articulated immobile objects¿ features. Experiments show that the proposed algorithm is faster than previous techniques and still achieves better accuracy.

In this paper, we propose to address the semantic- oriented 3D mesh hierarchical segmentation problem, using enhanced topological skeletons. This high level information drives both the feature boundary computation as well as the feature hierarchy definition. Proposed hierarchical scheme is based on the key idea that the topology of a feature is a more important decomposition criterion than its geometry. First, the enhanced topological skeleton of the input triangulated surface is constructed. Then it is used to delimit the core of the object and to identify junction areas. This second step results in a fine segmentation of the object. Finally, a fine to coarse strategy enables a semantic- oriented hierarchical composition of features, subdividing human limbs into arms and hands for example. Method performance is evaluated according to seven criteria enumerated in latest segmentation surveys [3]. Thanks to the high level description it uses as an input, presented approach results, with low computation times, in robust and meaningful compatible hierarchical decompositions.

In this paper, we propose a method for three-dimensional (3D)-model indexing based on two-dimensional (2D) views, which we call adaptive views clustering (AVC). The goal of this method is to provide an "optimal" selection of 2D views from a 3D model, and a probabilistic Bayesian method for 3D-model retrieval from these views. The characteristic view selection algorithm is based on an adaptive clustering algorithm and uses statistical model distribution scores to select the optimal number of views. Starting from the fact that all views do not have equal importance, we also introduce a novel Bayesian approach to improve the retrieval. Finally, we present our results and compare our method to some state-of-the-art 3D retrieval descriptors on the Princeton 3D Shape Benchmark database and a 3D-CAD-models database supplied by the car manufacturer Renault

In this paper, we use the three-dimensional topological shape information for human face identification. We propose a new method to represent 3D faces as a topological graph. Fine registration of surfaces is done by first automatically finding topological connected components, and then constructing its topological graph representing the important topological changes on the face. The similarity calculation between 3D faces is processed using coarse-to-fine strategy while preserving the consistency of the graph structures, which result in establishing a correspondence between the parts of faces. The experiments made with a 144 3D faces dataset show the efficiency of our approach