Generative motion prediction must satisfy three simultaneous requirements for real-world autonomy: high accuracy, diverse multimodal futures, and strictly bounded latency. Diffusion models meet the first two but violate the third, requiring tens to hundreds of denoising steps. We identify a conditioning strategy that resolves this tension: single-step integration is accurate when the underlying transport problem is local. A model that must both discover the correct behavioral mode and traverse a long displacement in one step accumulates large discretization errors; conditioning the base distribution to lie near plausible futures reduces the problem to short-range refinement, the regime where a single Euler step suffices. We instantiate this local transport conditioning in FlowS, a conditional flow matching framework with two mechanisms. First, an online, scene-conditioned learned prior emits K calibrated anchor trajectories per agent, each already near a plausible future, converting mode discovery into local correction. Second, a step-consistent displacement field enforces semigroup self-consistency, guaranteeing that a single step inherits multi-step accuracy. Crucially, anchoring this field at learned priors along straight-line paths yields a stable, low-variance training target, unlike prior self-consistency methods that suffer from high-variance bootstrap signals on curved diffusion paths. On the Waymo Open Motion Dataset, FlowS achieves state-of-the-art Soft mAP (0.4804) and mAP (0.4703) with ensemble at 75 FPS with single-step inference, demonstrating that local transport conditioning makes one-step generative motion prediction practical for safety-critical autonomy. Code and pretrained models will be released upon acceptance.

The evolution of Advanced Driver Assistance Systems (ADAS) has increased the need for robust and generalizable algorithms for multi-object tracking. Traditional statistical model-based tracking methods rely on predefined motion models and assumptions about system noise distributions. Although computationally efficient, they often lack adaptability to varying traffic scenarios and require extensive manual design and parameter tuning. To address these issues, we propose a novel 3D multi-object tracking approach for vehicles, HybridTrack, which integrates a data-driven Kalman Filter (KF) within a tracking-by-detection paradigm. In particular, it learns the transition residual and Kalman gain directly from data, which eliminates the need for manual motion and stochastic parameter modeling. Validated on the real-world KITTI dataset, HybridTrack achieves 82.72% HOTA accuracy, significantly outperforming state-of-the-art methods. We also evaluate our method under different configurations, achieving the fastest processing speed of 112 FPS. Consequently, HybridTrack eliminates the dependency on scene-specific designs while improving performance and maintaining real-time efficiency.

Residual error propagation remains a fundamental problem in recurrent models, where small prediction inaccuracies compound over time and degrade long-horizon performance. Accurately modeling the correlation structure of such residuals is critical for reliable uncertainty quantification in probabilistic multivariate timeseries forecasting. While recent time-series deep models efficiently parametrize time-varying contemporaneous correlations, they often assume temporal independence of errors and neglect spatial correlation across the observed network. In this paper, we introduce Teger, a structured uncertainty module that overcomes the spa- tial and temporal limitations of error-correlated autoregressive forecasting. Teger proposes a spatial curvature-aware graph rewiring mechanism explicitly strengthening information-bottleneck edges identified by discrete Forman curvature. The component is integrated into a low-rank-plus-diagonal covariance head, preserving tractable inference via the Woodbury identity. Teger is backbone-agnostic, requiring only the latent state produced by any autoregressive encoder. We provide theoretical evidence of Teger, and experimentally evaluate it on LSTM, Transformer, and xLSTM backbones across four real-world spatio-temporal datasets, showing consistent improvement in Continuous Ranked Probability Score (CRPS). We further provide a formal theoretical analysis connecting curvature-aware rewiring to (i) oversquashing alleviation, (ii) improved spectral connectivity, (iii) reduced effective resistance, and (iv) improved covariance calibration bounds

Time series data often suffers from resolution limitations due to hardware constraints, sampling frequency restrictions, or economic considerations. While super-resolution techniques have seen significant advancements in computer vision, their application to spatio-temporal data presents unique challenges that remain under-explored. We argue that pure generative or auto-regressive approaches are subpar for the multi-modal super-resolution task. Hence, we introduce ChronoFusion, a novel hybrid model that simultaneously enhances both spatial and temporal resolution of time series data. Our approach leverages a graph variational autoencoder combined with adaptive attention mechanisms to generate high-resolution time series from low-resolution inputs. Unlike previous methods that handle spatial and temporal super-resolution separately, ChronoFusion integrates both dimensions through a proxy subspace. Extensive evaluation on traffic datasets in various locations demonstrates that ChronoFusion outperforms state-of-the-art methods by 10% on average in interpolation fidelity on unseen nodes while maintaining temporal consistency. Furthermore, our model demonstrates strong capabilities in handling missing data. The method's versatility across diverse spatio-temporal traffic applications makes it a valuable contribution to time series analysis and modeling. Github Repo

This study explores the complementarity and fusion of two sensing technologies for pedestrian flow estimation: ubiquitous Wi-Fi-based passive radar (WPR) and deployable Frequency Modulated Continuous Wave (FMCW) active radar, both combined with a Convolutional Neural Network (CNN) for post processing. Wi-Fi signals, already widespread in many environments, enable passive, wide-area motion sensing without requiring additional infrastructure. FMCW radars, by contrast, offer high-resolution range-Doppler measurements and can be selectively deployed in target locations such as pedestrian streets. We begin by individually evaluating the performance of FMCW and Wi-Fi passive radar systems in estimating pedestrians flow, highlighting their respective strengths, limitations and complementarity. To further improve the system performance, we propose a fusion approach at both the decision and feature levels. For decision-level fusion, we implement majority voting and probability averaging strategies to combine the predictions from both radars. For feature-level fusion, we extract features from both radar systems using CNNs and merge them before classification. Our experimental results, from measurements collected on the same scene by the two radars, show that the fusion approaches significantly enhance the flow estimation accuracy compared to using either radar system alone. The feature-level fusion method, in particular, demonstrates superior performance by effectively integrating the spatial and motion information captured by both radar types. This work demonstrates the value of hybrid sensing systems that combine opportunistic and purpose-built technologies for reliable pedestrian counting and flow estimation in diverse urban scenarios and provides a robust framework for future developments in multi-sensor data fusion.

Multivariate time-series (MTS) forecasting is fundamental to applications ranging from urban mobility and resource management to climate modeling. While recent generative models based on denoising diffusion have advanced state-of-the-art performance in capturing complex data distributions, they suffer from significant computational overhead due to iterative stochastic sampling procedures that limit real-time deployment. Moreover, these models can be brittle when handling high-dimensional, non-stationary, and multi-scale periodic patterns characteristic of real-world sensor networks. We introduce FreqFlow, a novel framework that leverages conditional flow matching in the frequency domain for deterministic MTS forecasting. Unlike conventional approaches that operate in the time domain, FreqFlow transforms the forecasting problem into the spectral domain, where it learns to model amplitude and phase shifts through a single complex-valued linear layer. This frequency-domain formulation enables the model to efficiently capture temporal dynamics via complex multiplication, corresponding to scaling and temporal translations. The resulting architecture is exceptionally lightweight with only 89k parameters - an order of magnitude smaller than competing diffusion-based models-while enabling single-pass deterministic sampling through ordinary differential equation (ODE) integration. Our approach decomposes MTS signals into trend, seasonal, and residual components, with the flow matching mechanism specifically designed for residual learning to enhance long-term forecasting accuracy. Extensive experiments on real-world traffic speed, volume, and flow datasets demonstrate that FreqFlow achieves state-of-the-art forecasting performance, on average 7% RMSE improvements, while being significantly faster and more parameter-efficient than existing methods

In the past few years, there has been a rising demand for monocular vehicle 6D pose estimation in traffic video, due to its applicability in emerging domains such as smart mobility and intelligent transportation systems. However, most of the existing approaches are image-based, which results in undesirable temporal pose inconsistencies of vehicles between consecutive frames when applied to video sequences. In this work, we present a Kalman filter-based post-processing method, named MonoKalman, that enhances the temporal consistencies of the 6D pose estimates of vehicles in traffic video. We compare our method with a state-of-the-art 6D pose estimation method on synthetic video data of traffic scenery. The experimental results indicate that our MonoKalman significantly outperforms the image-based baseline method and effectively reduces temporal pose artifacts, ensuring a more coherent and stable representation of 6D vehicle temporal poses in traffic video. To more effectively demonstrate MonoKalman’s enhancements over the baseline model, we design a graphical user interface. This interface offers users insights through detailed quantitative metrics and dynamic visualizations, allowing them to conduct and customize their experiments.

Efficiently visualizing Spatial-Temporal traffic data plays an important role nowadays in traffic monitoring. Interactive dashboards offering effective visualizations of spatial-temporal traffic data play a more prominent role in traffic monitoring. In this paper, we introduce a dashboard for visualizing traffic data. Specifically, our dashboard integrates spatial-temporal components for the time-series traffic data of Brussels, which is the first GNN-based traffic demonstration tool for Brussels. Furthermore, we provide an interface for displaying traffic prediction of deep-learning-based Spatial-Temporal Graph Neural Networks (STGNNs), which have demonstrated state of the art performance in Intelligent Transpiration Systems (ITS). In addition, we demonstrate two real-world use cases by using the proposed dashboard which provides the potential for a future tool to achieve intelligent transportation management.

Traffic forecasting is pivotal for intelligent transportation systems, where accurate and interpretable predictions can significantly enhance operational efficiency and safety. A key challenge stems from the heterogeneity of traffic conditions across diverse locations, leading to highly varied traffic data distributions. Large language models (LLMs) show exceptional promise for few-shot learning in such dynamic and data-sparse scenarios. However, existing LLM-based solutions often rely on prompt-tuning, which can struggle to fully capture complex graph relationships and spatiotemporal dependencies-thereby limiting adaptability and interpretability in real-world traffic networks. We address these gaps by introducing Strada-LLM, a novel multivariate probabilistic forecasting LLM that explicitly models both temporal and spatial traffic patterns. By incorporating proximal traffic information as covariates, Strada-LLM more effectively captures local variations and outperforms prompt-based existing LLMs. To further enhance adaptability, we propose a lightweight distribution-derived strategy for domain adaptation, enabling parameter-efficient model updates when encountering new data distributions or altered network topologies-even under few-shot constraints. Empirical evaluations on spatio-temporal transportation datasets demonstrate that Strada-LLM consistently surpasses state-of-the-art LLM-driven and traditional GNN-based predictors. Specifically, it improves long-term forecasting by 17

A novel low-power distributed Visual Sensor Network (VSN) system is proposed, which performs real-time collaborative barcode localization, tracking, and robust identification. Due to a dynamic triggering mechanism and efficient transmission protocols, communication is organized amongst the nodes themselves rather than being orchestrated by a single sink node, achieving lower congestion and significantly reducing the vulnerability of the overall system. Specifically, early detection of the moving barcode is achieved through a dynamic triggering mechanism. A hierarchical transmission protocol is designed, within which different communication protocols are used, depending on the type of data exchanged among nodes. Real-Time Transport Protocol (RTP) is employed for video communication, while the Transmission Control Protocol (TCP) and Long Range (LoRa) protocol are used for passing messages amongst the nodes in the VSN. Through an extensive experimental evaluation, we demonstrate that the proposed distributed VSN brings substantial advantages in terms of accuracy, power savings, and time complexity compared to an equivalent system performing centralized processing.

In recent years, Vehicle Make and Model Recognition (VMMR) has attracted a lot of attention as it plays a crucial role in Intelligent Transportation Systems (ITS). Accurate and efficient VMMR systems are required in real-world applications including intelligent surveillance and autonomous driving. The paper introduces a new large-scale dataset and a novel deep learning paradigm for VMMR. A new large-scale dataset dubbed Diverse large-scale VMM (DVMM) is proposed collecting image-samples with the most popular vehicle brands operating in Europe. A novel VMMR framework is proposed which follows a two-branch architecture performing make and model recognition respectively. A two-stage training procedure and a novel decision module are proposed to process the make and model predictions and compute the final model prediction. In addition, a novel metric based on the true positive rate is proposed to compare classification confusion of the proposed 2B–2S and the baseline methods. A complex experimental validation is carried out, demonstrating the generality, diversity, and practicality of the proposed DVMM dataset. The experimental results show that the proposed framework provides 93.95% accuracy over the more diverse DVMM dataset and 95.85% accuracy over traditional VMMR datasets. The proposed two-branch approach outperforms the conventional one-branch approach for VMMR over small-, medium-, and large-scale datasets by providing lower vehicle model confusion and reduced inter-make ambiguity. The paper demonstrates the advantages of the proposed two-branch VMMR paradigm in terms of robustness and lower confusion relative to single-branch designs.

Multimodal, alias, guided, image restoration is the reconstruction of a degraded image from a target modality with the aid of a high quality image from another modality. A similar task is image fusion; it refers to merging images from different modalities into a composite image. Traditional approaches for multimodal image restoration and fusion include analytical methods that are computationally expensive at inference time. Recently developed deep learning methods have shown a great performance at a reduced computational cost; however, since these methods do not incorporate prior knowledge about the problem at hand, they result in a “black box” model, that is, one can hardly say what the model has learned. In this paper, we formulate multimodal image restoration and fusion as a coupled convolutional sparse coding problem, and adopt the Method of Multipliers (MM) for its solution. Then, we use the MM-based solution to design a convolutional neural network (CNN) encoder that follows the principle of deep unfolding. To address multimodal image restoration and fusion, we design two multimodal models which employ the proposed encoder followed by an appropriately designed decoder that maps the learned representations to the desired output. Unlike most existing deep learning designs comprising multiple encoding branches followed by a concatenation or a linear combination fusion block, the proposed design provides an efficient and structured way to fuse information at different stages of the network, providing representations that can lead to accurate image reconstruction. The proposed models are applied to three image restoration tasks, as well as two image fusion tasks. Quantitative and qualitative comparisons against various state-of-the-art analytical and deep learning methods corroborate the superior performance of the proposed framework.

It becomes common practice to equip our cities and streets with a wide range of sensors (e.g., cameras, inductive loops and push buttons for pedestrians). Those sensors are paving the way for smart monitoring of traffic, which is nowadays already used to, for example, dynamically steer traffic flows by controlling traffic lights or track stolen vehicles. However, such data remains very complex to thoroughly inspect and interpret. In this paper we exploit an advanced spatio-temporal disaggregation technique to extract insights from ANPR data. More specifically, we visualise and analyse spatial and temporal insights from vehicle detections along street segments in a dense network of ANPR cameras in the police district of Voorkempen, Belgium. The proposed analysis tools can be integrated in an intelligent transportation system (ITS) to support traffic monitoring by the appropriate authorities.

The latest advancements in the field of deep learning and biomedical engineering have allowed for the development of myoelectric interfaces based on deep neural networks. A longstanding problem of these interfaces is that the models cannot easily be applied to new users due to the high variability and stochastic nature of the electromyography signals. Further training a new model for every new subject requires the collection of large volumes of data. Therefore, this work proposes a transfer learning (TL) scheme which allows reusing the knowledge of a pre-existing model for a new user. Firstly, a convolutional neural network (CNN) is trained on an initial dataset using the data of multiple subjects. Then, the weights of this model are fine-tuned for a new target subject. The approach is evaluated on the Ninapro datasets DB2 and DB7. The experimentation included three different CNN models and eight preprocessing alternatives. The results showed that the success of the TL method depends on how the data are preprocessed. Specifically, the biggest accuracy improvement (+5.14%) is achieved when only the first 20% of the signal duration is used.

Multimodal alias, guided, image super-resolution (SR) refers to the reconstruction of a high-resolution (HR) version of a low-resolution (LR) image with the aid of an HR image from another image modality. Common approaches for the SR problem include analytical methods which are computationally expensive. Deep learning methods are capable of learning a nonlinear mapping between LR and HR images from data, delivering high reconstruction accuracy at a low-computational cost during inference; however, these methods do not incorporate any prior knowledge about the problem, with the neural network model behaving like a black box. In this paper, we formulate multimodal image SR as a coupled convolutional sparse coding problem. To solve the corresponding minimization problem, we adopt the Method of Multipliers (MM). We then design a convolutional neural network (CNN) that unfolds the obtained MM algorithm. The proposed CNN accepts as input the LR image from the main modality and the HR image from the guidance modality to reconstruct the desired HR image. Unlike existing deep learning methods, our CNN provides an efficient and structured way to fuse information at different stages of the network and achieves high reconstruction accuracy. We evaluate the performance of the proposed model for the super-resolution of multi-spectral images guided by their high resolution RGB counterparts.

Over the past few years, Deep learning (DL) has revolutionized the field of data analysis. Not only are the algorithmic paradigms changed, but also the performance in various classification and prediction tasks has been significantly improved with respect to the state-of-the-art, especially in the area of computer vision. The progress made in computer vision has produced a spillover in many other domains, such as biomedical engineering. Some recent works are directed towards surface electromyography (sEMG) based hand gesture recognition, often addressed as an image classification problem and solved using tools such as Convolutional Neural Networks (CNN). This paper extends our previous work on the application of the Hilbert space-filling curve for the generation of image representations from multi-electrode sEMG signals, by investigating how the Hilbert curve compares to the Peano- and Z-order space-filling curves. The proposed space-filling mapping methods are evaluated on a variety of network architectures and in some cases yield a classification improvement of at least 3%, when used to structure the inputs before feeding them into the original network architectures.

The paper proposes a novel instance segmentation method for traffic videos devised for deployment on real-time embedded devices. A novel neural network architecture is proposed using a multi-resolution feature extraction backbone and improved network designs for the object detection and instance segmentation branches. A novel post-processing method is introduced to ensure a reduced rate of false detection by evaluating the quality of the output masks. An improved network training procedure is proposed based on a novel label assignment algorithm. An ablation study on speed-vs.-performance trade-off further modifies the two branches and replaces the conventional ResNet-based performance-oriented backbone with a lightweight speed-oriented design. The proposed architectural variations achieve real-time performance when deployed on embedded devices. The experimental results demonstrate that the proposed instance segmentation method for traffic videos outperforms the you only look at coefficients algorithm, the state-of-the-art real-time instance segmentation method. The proposed architecture achieves qualitative results with 31.57 average precision on the COCO dataset, while its speed-oriented variations achieve speeds of up to 66.25 frames per second on the Jetson AGX Xavier module.

Museums all over the world store a large variety of digitized paintings and other works of art with significant historical value. Over time, these works of art deteriorate, making them lose their original splendour. For paintings, cracks and paint losses are the most prominent types of deterioration, mainly caused by environmental factors, such as fluctuations in temperature or humidity, improper storage conditions and even physical impacts. We propose a neural network architecture for the detection of crack patterns in paintings, using visual acquisitions from different modalities. The proposed architecture is composed of two neural network streams, one is a fully connected neural network while the other consists of a multiscale convolutional neural network. The convolutional neural network plays a leading role in the crack classification task, while the fully connected neural network plays an auxiliary role. To reduce the overall computational complexity of the proposed method, we use morphological filtering as a pre-processing step to safely exclude areas of the image that do not contain cracks and do not need further processing. We validate the proposed method on a multimodal visual dataset from the Ghent Altarpiece, a world famous polyptych by the Van Eyck brothers. The results show an encouraging performance of the proposed approach compared to traditional machine learning methods and the state-of-the-art Bayesian Conditional Tensor Factorization (BCTF) method for crack detection.

The reconstruction of a high resolution image given a low resolution observation is an ill-posed inverse problem in imaging. Deep learning methods rely on training data to learn an end-to-end mapping from a low-resolution input to a highresolution output. Unlike existing deep multimodal models that do not incorporate domain knowledge about the problem, we propose a multimodal deep learning design that incorporates sparse priors and allows the effective integration of information from another image modality into the network architecture. Our solution relies on a novel deep unfolding operator, performing steps similar to an iterative algorithm for convolutional sparse coding with side information; therefore, the proposed neural network is interpretable by design. The deep unfolding architecture is used as a core component of a multimodal framework for guided image super-resolution. An alternative multimodal design is investigated by employing residual learning to improve the training efficiency. The presented multimodal approach is applied to super-resolution of near-infrared and multi-spectral images as well as depth upsampling using RGB images as side information. Experimental results show that our model outperforms state-ofthe-art methods.