arXiv subjects

Ahmed Helmy

Publications and source records attributed to Ahmed Helmy.

At least 19 recordsLinked to original sources

arXiv2026

LiveGesture Streamable Co-Speech Gesture Generation Model

We propose LiveGesture, the first fully streamable, speech-driven full-body gesture generation framework that operates with zero look-ahead and supports arbitrary sequence length. Unlike existing co-speech gesture methods, which are designed for offline generation and either treat body regions independently or entangle all joints within a single model, LiveGesture is built from the ground up for causal, region-coordinated motion generation. LiveGesture consists of two main modules: the Streamable Vector Quantized Motion Tokenizer (SVQ) and the Hierarchical Autoregressive Transformer (HAR). The SVQ tokenizer converts the motion sequence of each body region into causal, discrete motion tokens, enabling real-time, streamable token decoding. On top of SVQ, HAR employs region-expert autoregressive (xAR) transformers to model expressive, fine-grained motion dynamics for each body region. A causal spatio-temporal fusion module (xAR Fusion) then captures and integrates correlated motion dynamics across regions. Both xAR and xAR Fusion are conditioned on live, continuously arriving audio signals encoded by a streamable causal audio encoder. To enhance robustness under streaming noise and prediction errors, we introduce autoregressive masking training, which leverages uncertainty-guided token masking and random region masking to expose the model to imperfect, partially erroneous histories during training. Experiments on the BEAT2 dataset demonstrate that LiveGesture produces coherent, diverse, and beat-synchronous full-body gestures in real time, matching or surpassing state-of-the-art offline methods under true zero look-ahead conditions.

cs.CV↗

arXiv2026

Monocular Models are Strong Learners for Multi-View Human Mesh Recovery

Multi-view human mesh recovery (HMR) is broadly deployed in diverse domains where high accuracy and strong generalization are essential. Existing approaches can be broadly grouped into geometry-based and learning-based methods. However, geometry-based methods (e.g., triangulation) rely on cumbersome camera calibration, while learning-based approaches often generalize poorly to unseen camera configurations due to the lack of multi-view training data, limiting their performance in real-world scenarios. To enable calibration-free reconstruction that generalizes to arbitrary camera setups, we propose a training-free framework that leverages pretrained single-view HMR models as strong priors, eliminating the need for multi-view training data. Our method first constructs a robust and consistent multi-view initialization from single-view predictions, and then refines it via test-time optimization guided by multi-view consistency and anatomical constraints. Extensive experiments demonstrate state-of-the-art performance on standard benchmarks, surpassing multi-view models trained with explicit multi-view supervision.

cs.CV↗

arXiv2025

MonoMSK: Monocular 3D Musculoskeletal Dynamics Estimation

Reconstructing biomechanically realistic 3D human motion - recovering both kinematics (motion) and kinetics (forces) - is a critical challenge. While marker-based systems are lab-bound and slow, popular monocular methods use oversimplified, anatomically inaccurate models (e.g., SMPL) and ignore physics, fundamentally limiting their biomechanical fidelity. In this work, we introduce MonoMSK, a hybrid framework that bridges data-driven learning and physics-based simulation for biomechanically realistic 3D human motion estimation from monocular video. MonoMSK jointly recovers both kinematics (motions) and kinetics (forces and torques) through an anatomically accurate musculoskeletal model. By integrating transformer-based inverse dynamics with differentiable forward kinematics and dynamics layers governed by ODE-based simulation, MonoMSK establishes a physics-regulated inverse-forward loop that enforces biomechanical causality and physical plausibility. A novel forward-inverse consistency loss further aligns motion reconstruction with the underlying kinetic reasoning. Experiments on BML-MoVi, BEDLAM, and OpenCap show that MonoMSK significantly outperforms state-of-the-art methods in kinematic accuracy, while for the first time enabling precise monocular kinetics estimation.

cs.CV↗

arXiv2025

Walk Before You Dance: High-fidelity and Editable Dance Synthesis via Generative Masked Motion Prior

Recent advances in dance generation have enabled the automatic synthesis of 3D dance motions. However, existing methods still face significant challenges in simultaneously achieving high realism, precise dance-music synchronization, diverse motion expression, and physical plausibility. To address these limitations, we propose a novel approach that leverages a generative masked text-to-motion model as a distribution prior to learn a probabilistic mapping from diverse guidance signals, including music, genre, and pose, into high-quality dance motion sequences. Our framework also supports semantic motion editing, such as motion inpainting and body part modification. Specifically, we introduce a multi-tower masked motion model that integrates a text-conditioned masked motion backbone with two parallel, modality-specific branches: a music-guidance tower and a pose-guidance tower. The model is trained using synchronized and progressive masked training, which allows effective infusion of the pretrained text-to-motion prior into the dance synthesis process while enabling each guidance branch to optimize independently through its own loss function, mitigating gradient interference. During inference, we introduce classifier-free logits guidance and pose-guided token optimization to strengthen the influence of music, genre, and pose signals. Extensive experiments demonstrate that our method sets a new state of the art in dance generation, significantly advancing both the quality and editability over existing approaches. Project Page available at https://foram-s1.github.io/DanceMosaic/

cs.GR↗

arXiv2025

MaskHand: Generative Masked Modeling for Robust Hand Mesh Reconstruction in the Wild

Reconstructing a 3D hand mesh from a single RGB image is challenging due to complex articulations, self-occlusions, and depth ambiguities. Traditional discriminative methods, which learn a deterministic mapping from a 2D image to a single 3D mesh, often struggle with the inherent ambiguities in 2D-to-3D mapping. To address this challenge, we propose MaskHand, a novel generative masked model for hand mesh recovery that synthesizes plausible 3D hand meshes by learning and sampling from the probabilistic distribution of the ambiguous 2D-to-3D mapping process. MaskHand consists of two key components: (1) a VQ-MANO, which encodes 3D hand articulations as discrete pose tokens in a latent space, and (2) a Context-Guided Masked Transformer that randomly masks out pose tokens and learns their joint distribution, conditioned on corrupted token sequence, image context, and 2D pose cues. This learned distribution facilitates confidence-guided sampling during inference, producing mesh reconstructions with low uncertainty and high precision. Extensive evaluations on benchmark and real-world datasets demonstrate that MaskHand achieves state-of-the-art accuracy, robustness, and realism in 3D hand mesh reconstruction. Project website: https://m-usamasaleem.github.io/publication/MaskHand/MaskHand.html.

cs.CV↗

arXiv2025

BioPose: Biomechanically-accurate 3D Pose Estimation from Monocular Videos

Recent advancements in 3D human pose estimation from single-camera images and videos have relied on parametric models, like SMPL. However, these models oversimplify anatomical structures, limiting their accuracy in capturing true joint locations and movements, which reduces their applicability in biomechanics, healthcare, and robotics. Biomechanically accurate pose estimation, on the other hand, typically requires costly marker-based motion capture systems and optimization techniques in specialized labs. To bridge this gap, we propose BioPose, a novel learning-based framework for predicting biomechanically accurate 3D human pose directly from monocular videos. BioPose includes three key components: a Multi-Query Human Mesh Recovery model (MQ-HMR), a Neural Inverse Kinematics (NeurIK) model, and a 2D-informed pose refinement technique. MQ-HMR leverages a multi-query deformable transformer to extract multi-scale fine-grained image features, enabling precise human mesh recovery. NeurIK treats the mesh vertices as virtual markers, applying a spatial-temporal network to regress biomechanically accurate 3D poses under anatomical constraints. To further improve 3D pose estimations, a 2D-informed refinement step optimizes the query tokens during inference by aligning the 3D structure with 2D pose observations. Experiments on benchmark datasets demonstrate that BioPose significantly outperforms state-of-the-art methods. Project website: \url{https://m-usamasaleem.github.io/publication/BioPose/BioPose.html}.

cs.CV↗

arXiv2025

MS-Temba: Multi-Scale Temporal Mamba for Understanding Long Untrimmed Videos

Temporal Action Detection (TAD) in untrimmed videos poses significant challenges, particularly for Activities of Daily Living (ADL) requiring models to (1) process long-duration videos, (2) capture temporal variations in actions, and (3) simultaneously detect dense overlapping actions. Existing CNN and Transformer-based approaches, struggle to jointly capture fine-grained detail and long-range structure at scale. State-space Model (SSM) based Mamba offers powerful long-range modeling, but naive application to TAD collapses fine-grained temporal structure and fails to account for the challenges inherent to TAD. To this end, we propose Multi-Scale Temporal Mamba (MS-Temba), which extends Mamba to TAD with newly introduced dilated SSMs. Each Temba block, comprising dilated SSMs coupled with our proposed additional losses, enables the learning of discriminative representations across temporal scales. A lightweight Multi-scale Mamba Fuser then unifies these multi-scale features via SSM-based aggregation, yielding precise action-boundary localization. With only 17M parameters, MS-Temba achieves state-of-the-art performance on densely labeled ADL benchmarks TSU & Charades, and further generalizes to long-form video summarization, setting new state-of-the-art results on TVSum & SumMe.

cs.CV↗

arXiv2020

Analyzing Mobility-Traffic Correlations in Large WLAN Traces: Flutes vs. Cellos

Two major factors affecting mobile network performance are mobility and traffic patterns. Simulations and analytical-based performance evaluations rely on models to approximate factors affecting the network. Hence, the understanding of mobility and traffic is imperative to the effective evaluation and efficient design of future mobile networks. Current models target either mobility or traffic, but do not capture their interplay. Many trace-based mobility models have largely used pre-smartphone datasets (e.g., AP-logs), or much coarser granularity (e.g., cell-towers) traces. This raises questions regarding the relevance of existing models, and motivates our study to revisit this area. In this study, we conduct a multidimensional analysis, to quantitatively characterize mobility and traffic spatio-temporal patterns, for laptops and smartphones, leading to a detailed integrated mobility-traffic analysis. Our study is data-driven, as we collect and mine capacious datasets (with 30TB, 300k devices) that capture all of these dimensions. The investigation is performed using our systematic (FLAMeS) framework. Overall, dozens of mobility and traffic features have been analyzed. The insights and lessons learnt serve as guidelines and a first step towards future integrated mobility-traffic models. In addition, our work acts as a stepping-stone towards a richer, more-realistic suite of mobile test scenarios and benchmarks.

cs.NI↗

arXiv2019

Practical Prediction of Human Movements Across Device Types and Spatiotemporal Granularities

Understanding and predicting mobility are essential for the design and evaluation of future mobile edge caching and networking. Consequently, research on prediction of human mobility has drawn significant attention in the last decade. Employing information-theoretic concepts and machine learning methods, earlier research has shown evidence that human behavior can be highly predictable. Despite existing studies, more investigations are needed to capture intrinsic mobility characteristics constraining predictability, and to explore more dimensions (e.g. device types) and spatio-temporal granularities, especially with the change in human behavior and technology. We analyze extensive longitudinal datasets with fine spatial granularity (AP level) covering 16 months. The study reveals device type as an important factor affecting predictability. Ultra-portable devices such as smartphones have "on-the-go" mode of usage (and hence dubbed "Flutes"), whereas laptops are "sit-to-use" (dubbed "Cellos"). The goal of this study is to investigate practical prediction mechanisms to quantify predictability as an aspect of human mobility modeling, across time, space and device types. We apply our systematic analysis to wireless traces from a large university campus. We compare several algorithms using varying degrees of temporal and spatial granularity for the two modes of devices; Flutes vs. Cellos. Through our analysis, we quantify how the mobility of Flutes is less predictable than the mobility of Cellos. In addition, this pattern is consistent across various spatio-temporal granularities, and for different methods (Markov chains, neural networks/deep learning, entropy-based estimators). This work substantiates the importance of predictability as an essential aspect of human mobility, with direct application in predictive caching, user behavior modeling and mobility simulations.

cs.NI↗

arXiv2018

Interest-Aware Delivery for Mobile Social Networks: A TRACE-driven Approach

We envision future mobile networks to be human-centric supporting interest-aware delivery, where an interest maybe based on behavior, such as mobility pattern, location, or web browsing (or user profile) such as affiliation, attributes, or activity. An essential capability in future networks will be the ability to provide scalable group communication. Current communication paradigms, including unicast and multicast, require explicit identification of destination nodes (through node IDs or group membership protocols), while directory services translate logical, interest-specific queries into destination IDs where parties are then connected using interest-oblivious protocols. The power and scalability of such paradigms is quite limited in the context of future, highly dynamic mobile social networks, where it is desirable in many scenarios to support implicit membership based on interest. In such scenarios, membership in interest-groups is not explicitly expressed by users, it is rather autonomously inferred by network protocols based on behavioral profiles. This removes the dependence on third parties (via directory lookup), the need for explicit expression of interest (or human intervention), and minimizes delivery overhead to uninterested users. While existing types of social networks will likely evolve in the context of mobile networks, we provide a novel paradigm of communication, deviating significantly from existing approaches and subsequently enabling qualitatively different capabilities for future services in mobile societies. In addition, our services and protocol design philosophy begins by analyzing the context in which the service is to be deployed, unlike most existing wireless networking approaches of designing general purpose protocols.

cs.SI↗

arXiv2018

Playing with Matches: Vehicular Mobility through Analysis of Trip Similarity and Matching

Understanding city-scale vehicular mobility and trip patterns is essential to addressing many problems, from transportation and pollution to public safety, among others. Using spatio-temporal analysis of vehicular mobility, promising solutions can be proposed to alleviate these major challenges, utilizing shared mobility and crowd-sourcing. The rise of transportation networks (e.g. Uber, Lyft), is a mere beginning to shared mobility. In this paper, we address problems of trip representation and matching. Particularly, we study a real-world dataset of trips (from Cologne, Germany), from spatial and temporal perspectives. Comparison of trajectories is desired for applications relying on spatio-temporal phenomena. For that purpose, we present a novel combined spatio-temporal similarity score, based on the weighted geometric mean (WGM) and conduct experiments on its applicability and strengths. First, we use the score to find clusters of trips that were spatially and/or temporally separable using spectral clustering. The score is then used in a real-time matching of trips for Catch-a-Ride (CaR) and CarPooling (CP) scenarios. CaR and CP achieve $\approx40\%$ and $\approx25\%$ decrease in traveled distances respectively, at the cost of moving to pick-up and from drop-off locations (i.e. drivers going on average $<700m$ out of their way on pick-up and drop-off for CP). Additionally, a comparison with the metrics available in the literature is presented on CaR scenario. We find that main advantages of WGM include the flexibility to favor time or space components, and linearity of runtime complexity. Finally, we formulate an optimal free float Car-Sharing scenario (e.g. scheduling a system of automated vehicles or taxis) resulting in an average of $\approx3.88$ trips serviced by a car in one hour.

cs.CE↗

arXiv2018

On the Relation Between Mobile Encounters and Web Traffic Patterns: A Data-driven Study

Mobility and network traffic have been traditionally studied separately. Their interaction is vital for generations of future mobile services and effective caching, but has not been studied in depth with real-world big data. In this paper, we characterize mobility encounters and study the correlation between encounters and web traffic profiles using large-scale datasets (30TB in size) of WiFi and NetFlow traces. The analysis quantifies these correlations for the first time, across spatio-temporal dimensions, for device types grouped into on-the-go Flutes and sit-to-use Cellos. The results consistently show a clear relation between mobility encounters and traffic across different buildings over multiple days, with encountered pairs showing higher traffic similarity than non-encountered pairs, and long encounters being associated with the highest similarity. We also investigate the feasibility of learning encounters through web traffic profiles, with implications for dissemination protocols, and contact tracing. This provides a compelling case to integrate both mobility and web traffic dimensions in future models, not only at an individual level, but also at pairwise and collective levels. We have released samples of code and data used in this study on GitHub, to support reproducibility and encourage further research (https://github.com/BabakAp/encounter-traffic).

cs.NI↗

arXiv2016

Multi-modal Mining and Modeling of Big Mobile Networks Based on Users Behavior and Interest

Usage of mobile wireless Internet has grown very fast in recent years. This radical change in availability of Internet has led to communication of big amount of data over mobile networks and consequently new challenges and opportunities for modeling of mobile Internet characteristics. While the traditional approach toward network modeling suggests finding a generic traffic model for the whole network, in this paper, we show that this approach does not capture all the dynamics of big mobile networks and does not provide enough accuracy. Our case study based on a big dataset including billions of netflow records collected from a campus-wide wireless mobile network shows that user interests acquired based on accessed domains and visited locations as well as user behavioral groups have a significant impact on traffic characteristics of big mobile networks. For this purpose, we utilize a novel graph-based approach based on KS-test as well as a novel co-clustering technique. Our study shows that interest-based modeling of big mobile networks can significantly improve the accuracy and reduce the KS distance by factor of 5 comparing to the generic approach.

cs.NI↗

arXiv2015

An Information-theoretic Model for Knowledge Sharing in Opportunistic Social Networks

In this paper we establish fundamental limits on the performance of knowledge sharing in opportunistic social net- works. In particular, we introduce a novel information-theoretic model to characterize the performance limits of knowledge sharing policies. Towards this objective, we first introduce the notions of knowledge gain and its upper bound, knowledge gain limit, per user. Second, we characterize these quantities for a number of network topologies and sharing policies. This work constitutes a first step towards defining and characterizing the performance limits and trade-offs associated with knowledge sharing in opportunistic social networks. Finally, we present nu- merical results characterizing the cumulative knowledge gain over time and its upper bound, using publicly available smartphone data. The results confirm the key role of the proposed model to motivate future research in this ripe area of research as well as new knowledge sharing policies.

cs.NI↗

arXiv2012

Comparing Background Subtraction Algorithms and Method of Car Counting

In this paper, we compare various image background subtraction algorithms with the ground truth of cars counted. We have given a sample of thousand images, which are the snap shots of current traffic as records at various intersections and highways. We have also counted an approximate number of cars that are visible in these images. In order to ascertain the accuracy of algorithms to be used for the processing of million images, we compare them on many metrics that includes (i) Scalability (ii) Accuracy (iii) Processing time.

cs.CV↗

arXiv2011

Towards Realistic Vehicular Network Modeling Using Planet-scale Public Webcams

Realistic modeling of vehicular mobility has been particularly challenging due to a lack of large libraries of measurements in the research community. In this paper we introduce a novel method for large-scale monitoring, analysis, and identification of spatio-temporal models for vehicular mobility using the freely available online webcams in cities across the globe. We collect vehicular mobility traces from 2,700 traffic webcams in 10 different cities for several months and generate a mobility dataset of 7.5 Terabytes consisting of 125 million of images. To the best of our knowl- edge, this is the largest data set ever used in such study. To process and analyze this data, we propose an efficient and scalable algorithm to estimate traffic density based on background image subtraction. Initial results show that at least 82% of individual cameras with less than 5% deviation from four cities follow Loglogistic distribution and also 94% cameras from Toronto follow gamma distribution. The aggregate results from each city also demonstrate that Log- Logistic and gamma distribution pass the KS-test with 95% confidence. Furthermore, many of the camera traces exhibit long range dependence, with self-similarity evident in the aggregates of traffic (per city). We believe our novel data collection method and dataset provide a much needed contribution to the research community for realistic modeling of vehicular networks and mobility.

cs.NI↗

arXiv2010

Mobile Testbeds with an Attitude

There have been significant recent advances in mobile networks, specifically in multi-hop wireless networks including DTNs and sensor networks. It is critical to have a testing environment to realistically evaluate such networks and their protocols and services. Towards this goal, we propose a novel, mobile testbed of two main components. The first consists of a network of robots with personality- mimicking, human-encounter behaviors, which will be the focus of this demo. The personality is build upon behavioral profiling of mobile users based on extensive wireless-network measurements and analysis. The second component combines the testbed with the human society using a new concept that we refer to as participatory testing utilizing crowd sourcing.

cs.NI↗

arXiv2010

Spatio-Temporal Modeling of Wireless Users Internet Access Patterns Using Self-Organizing Maps

User online behavior and interests will play a central role in future mobile networks. We introduce a systematic method for large-scale multi-dimensional analysis of online activity for thousands of mobile users across 79 buildings over a variety of web domains. We propose a modeling approach based on self-organizing maps (SOM) for discovering, organizing and visualizing different mobile users' trends from billions of WLAN records. We find surprisingly that users' trends based on domains and locations can be accurately modeled using a self-organizing map with clearly distinct characteristics. We also find many non-trivial correlations between different types of web domains and locations. Based on our analysis, we introduce a mixture model as an initial step towards realistic simulation of wireless network usage.

cs.NI↗