Md. Zasim Uddin

Dr. Md. Zasim Uddin

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 October 2024 in IEEE Access
Verified
Journal Article
Open access
Kamrul HasanKamrul Hasan
Kamrul Hasan
MUMd. Zasim Uddin
Md. Zasim Uddin
ARAusrukona Ray
Ausrukona Ray
Mahmudul HasanMahmudul Hasan
Mahmudul Hasan
FAFady Alnajjar
Fady Alnajjar
M.A.R. AhadM.A.R. Ahad
M.A.R. Ahad
Gait recognition is an advanced biometric technology that can be used to identify individuals based on their walking patterns, even from low-spatial-resolution image sequences from security surveillance camera footage. Traditional gait recognition approaches rely on complete body information and often overlook the challenge of occlusion. In real-world scenarios, various body parts may be occluded by physical obstacles such as buildings, walls, fences, vehicles, trees, or even other individuals in crowded areas. This occlusion results in a significant portion of the human body being unobserved, causing conventional gait recognition approaches to fail to identify the person. To address this challenge, we have developed a novel framework for gait recognition in the presence of occlusion, incorporating occlusion detection and reconstruction (ODR) and feature extraction for gait recognition (FEGR) modules. The ODR module identifies the occlusion type and reconstructs the occluded portions of the human body in a silhouette sequence using three-dimensional (3D) generative adversarial networks, whereas the FEGR module extracts partwise global and local features using 3D convolutional neural networks (CNNs) and full body features on a frame-by-frame basis using two-dimensional CNNs. We validated our framework using the CASIA-B and OU-MVLP datasets with artificially added occlusions and found that it showed superior performance, with average rank-1 accuracies of 96.4%, 87.8%, and 69.2% for normal, carried object, and clothing variations on CASIA-B and 58.9% on OU-MVLP, as well as 100.0% occlusion detection accuracy. These results demonstrate the ability of our proposed framework to maintain superior gait recognition performance despite the presence of occlusions.
1

30 November 2024 in Sensors
Verified
Journal Article
Open access
PDF file
ARAusrukona Ray
Ausrukona Ray
MUMd. Zasim Uddin
Md. Zasim Uddin
Kamrul HasanKamrul Hasan
Kamrul Hasan
ZMZinat Rahman Melody
Zinat Rahman Melody
PSProdip Kumar Sarker
Prodip Kumar Sarker
M.A.R. AhadM.A.R. Ahad
M.A.R. Ahad
Gait recognition is a behavioral biometric technique that identifies individuals based on their unique walking patterns, enabling long-distance identification. Traditional gait recognition methods rely on appearance-based approaches that utilize background-subtracted silhouette sequences to extract gait features. While effective and easy to compute, these methods are susceptible to variations in clothing, carried objects, and illumination changes, compromising the extraction of discriminative features in real-world applications. In contrast, model-based approaches using skeletal key points offer robustness against these covariates. Advances in human pose estimation (HPE) algorithms using convolutional neural networks (CNNs) have facilitated the extraction of skeletal key points, addressing some challenges of model-based approaches. However, the performance of skeleton-based methods still lags behind that of appearance-based approaches. This paper aims to bridge this performance gap by introducing a multi-biometric framework that extracts features from multiple HPE algorithms for gait recognition, employing feature-level fusion (FLF) and decision-level fusion (DLF) by leveraging a single-source multi-sample technique. We utilized state-of-the-art HPE algorithms, OpenPose, AlphaPose, and HRNet, to generate diverse skeleton data samples from a single source video. Subsequently, we employed a residual graph convolutional network (ResGCN) to extract features from the generated skeleton data. In the FLF approach, the features extracted from ResGCN and applied to the skeleton data samples generated by multiple HPE algorithms are aggregated point-wise for gait recognition, while in the DLF approach, the decisions of ResGCN applied to each skeleton data sample are integrated using majority voting for the final recognition. Our proposed method demonstrated state-of-the-art skeleton-based cross-view gait recognition performance on a popular dataset, CASIA-B.

9 December 2024 in IEEE Access
Verified
Journal Article
Open access
MUMd. Zasim Uddin
Md. Zasim Uddin
Kamrul HasanKamrul Hasan
Kamrul Hasan
M.A.R. AhadM.A.R. Ahad
M.A.R. Ahad
FAFady Alnajjar
Fady Alnajjar
Gait recognition, a form of biometric authentication, facilitates the identification of individuals by analyzing their characteristic walking patterns. This approach exhibits superior performance even from distant, low-resolution imagery from security camera footage. Historically, gait recognition methodologies used the entire sequence of a human body silhouette for spatiotemporal characterization. Recent advancements have introduced part-based feature extraction modules derived from the human body’s transverse plane (i.e., horizontal direction) into cross-view gait recognition (CVGR) applications. However, this study reveals the considerable potential of the parts in the sagittal plane (i.e., vertical direction) to enhance discrimination in CVGR. A novel method is proposed that integrates the parts generated according to transverse and sagittal planes utilizing three-dimensional and two-dimensional convolutional neural networks for robust feature extraction. The proposed method comprises a global, horizontal, and vertical part module for capturing fine-grained local details in the horizontal and vertical part directions, and a horizontal and vertical pyramid mapping module for extracting spatial features into the horizontal and vertical pyramid mapping. The consolidated features from both modules enhance CVGR performance, even amidst challenging covariates such as different carried objects and clothing variations, along with uncontrolled walking patterns in the wild. The effectiveness of this method is demonstrated through its implementation on the CASIA-B, OU-MVLP, and Gait3D benchmark datasets, where it exhibits superior gait recognition performance.

20 August 2025 in IEEE Transactions on Human-Machine Systems
Verified
Journal Article
Closed access
MUMd. Zasim Uddin
Md. Zasim Uddin
ARAusrukona Ray
Ausrukona Ray
BDBorsha Das
Borsha Das
M.A.R. AhadM.A.R. Ahad
M.A.R. Ahad
Gait has emerged as a promising biometric modality due to its noninvasive nature and the ability to capture samples from a distance. Model-based gait recognition using skeleton data conveys rich information that remains invariant to carried objects and clothing variations. However, viewing a person from different angles alters their gait posture, resulting in increased intrasubject variability compared to intersubject variability. Therefore, we propose a novel framework, view-embedding modified residual graph convolutional network (VeMResGCN), for cross-view gait recognition (CVGR) by exploiting two modules: modified residual graph convolutional network (MResGCN) and view-embedding feature extraction (VeFE) for view-invariant features. A state-of-the-art pose estimation algorithm extracts skeleton key points from raw video input, from which multiple features (e.g., relative joint positions, motion velocities, and bone structures) are computed. The final feature vector for gait recognition is computed by consolidating the features from the MResGCN and VeFE modules. To the best of authors’ knowledge, this work is the first to extract view-invariant features in a unified graph convolutional network (GCN) for skeleton-based CVGR. We evaluate our proposed framework on two of the largest publicly available skeleton datasets, CASIA-B and OUMVLP-Pose, under challenging covariates of clothing variation and carried objects. Results demonstrate that VeMResGCN significantly outperforms state-of-the-art methods with average rank-1 accuracies of 90.3%, 80.7%, and 73.4% for normal, carried object, and clothing variations on CASIA-B, and 71.0% on OU-MVLP in terms of skeleton-based CVGR. These results demonstrate the ability of our proposed framework to maintain superior CVGR performance despite the presence of carried objects and clothing variations. The proposed framework holds strong implications for real-world biometric applications, including robust person reidentification and surveillance systems, where maintaining consistent recognition across varying views and covariates is crucial.

19 July 2025 in Symmetry
Verified
Journal Article
Open access
PDF file
Kamrul HasanKamrul Hasan
Kamrul Hasan
KTKhandokar Alisha Tuhin
Khandokar Alisha Tuhin
Md Rasul Islam BaparyMd Rasul Islam Bapary
Md Rasul Islam Bapary
SDShafi Ud Doula
Shafi Ud Doula
Md Ashraful AlamMd Ashraful Alam
Md Ashraful Alam
MUMd. Zasim Uddin
Md. Zasim Uddin
Gait recognition is a reliable biometric approach that uniquely identifies individuals based on their natural walking patterns. It is widely used to recognize individuals who are challenging to camouflage and do not require a person’s cooperation. The general face-based person recognition system often fails to determine the offender’s identity when they conceal their face by wearing helmets and masks to evade identification. In such cases, gait-based recognition is ideal for identifying offenders, and most existing work leverages a deep learning (DL) model. However, a single model often fails to capture a comprehensive selection of refined patterns in input data when external factors are present, such as variation in viewing angle, clothing, and carrying conditions. In response to this, this paper introduces a fusion-based multi-model gait recognition framework that leverages the potential of convolutional neural networks (CNNs) and a vision transformer (ViT) in an ensemble manner to enhance gait recognition performance. Here, CNNs capture spatiotemporal features, and ViT features multiple attention layers that focus on a particular region of the gait image. The first step in this framework is to obtain the Gait Energy Image (GEI) by averaging a height-normalized gait silhouette sequence over a gait cycle, which can handle the left–right gait symmetry of the gait. After that, the GEI image is fed through multiple pre-trained models and fine-tuned precisely to extract the depth spatiotemporal feature. Later, three separate fusion strategies are conducted, and the first one is decision-level fusion (DLF), which takes each model’s decision and employs majority voting for the final decision. The second is feature-level fusion (FLF), which combines the features from individual models through pointwise addition before performing gait recognition. Finally, a hybrid fusion combines DLF and FLF for gait recognition. The performance of the multi-model fusion-based framework was evaluated on three publicly available gait databases: CASIA-B, OU-ISIR D, and the OU-ISIR Large Population dataset. The experimental results demonstrate that the fusion-enhanced framework achieves superior performance.

17 October 2025 in Frontiers in Digital Health
Verified
Journal Article
Open access
PDF file
MUMd. Zasim Uddin
Md. Zasim Uddin
ASArif Shahriar
Arif Shahriar
BSBjörn W. Schuller
Björn W. Schuller
NMNadim Mahamood
Nadim Mahamood
AAAtiqur Rahman Ahad
Atiqur Rahman Ahad
Introduction: Early skin disease diagnosis is essential and one of the challenging tasks for a dermatologist. Manual diagnosis by healthcare providers is subjective, costly, and may yield inconsistent results. In contrast, automated skin disease detection and classification using traditional machine learning and deep learning approaches have shown promise in addressing this problem.Methods: In this study, we propose a hybrid ensemble framework that integrates both feature-level fusion (FLF) and decision-level fusion (DLF) to leverage complementary strengths for detecting and classifying skin diseases. We employ two convolutional neural network (CNN)-based models, i.e., a modified DenseNet201 and VGG19, along with an attention-based model vision transformer (ViT) to identify and classify skin diseases. In FLF, feature representations from these models are point-wise added and passed through a shared classification head to make the final prediction. In DLF, decisions from each base model are collected, and the majority voting scheme is used to make a final decision. Furthermore, we incorporate a generative adversarial network (GAN)-based approach for offline-based training data augmentation to reduce overfitting and improve performance.Results: Based on different evaluation metrics (i.e., accuracy, precision, recall, and F1-score), our proposed framework demonstrates superior performance on four benchmark datasets: the PH2, HAM10000, ISIC 2018, and ISIC 2019 datasets, with an accuracy of 99.3%/99.2%, 92.7%/96.1%, 86.7%/89.0%, and 94.5%/95.0%, respectively, for FLF/DLF.Discussion: These results demonstrate that while both fusion strategies are effective, DLF slightly outperforms FLF, emphasizing the value of ensemble decision aggregation for robust skin disease classification.