I am a doctoral researcher in the Deep Learning Lab at West Virginia University, working with Prof. Nasser Nasrabadi. My current research focuses on developing machine learning and deep learning algorithms for applications in computer vision and biometrics. I have over 7 years of research experience and more than 2 years of teaching experience. I have published 14+ papers in renowned international computer vision and biometrics conferences such as WACV, IJCB, and BIOSIG. Additionally, I have journal publications in IEEE Transactions on Aerospace and Electronic Systems, as well as IET Biometrics, among others.
Prior to joining WVU, I served as a Lecturer in the Department of Computer Science at Manarat International University. I completed my M.Sc. at IICT from Bangladesh University of Engineering & Technology (BUET), under the guidance of Prof. Hossen Asiful Mustafa. I obtained my B.Sc. at the department of EEE from Khulna University of Engineering and Technology (KUET) under the supervision of Dr. Mohiuddin Ahmad.
If you have any questions of my work or seeking any form of academic cooperation, please feel free to email me at mahedi0803@gmail.com.
🔥 News
- 2024.01: 🔥 My google scholar citations have exceeded 200!
- 2023.12: 🎉 Our journal on ATR is accepted by IEEE Transactions on Aerospace and Electronic Systems.
- 2023.10: 🎉 Our paper on Face Recognition is accepted by WACV 2024 code video
- 2023.06: 🎉 One paper Caption-guided face recognition is accepted by IJCB 2023 code
- 2023.04: Attended EAB and CITeR Biometrics Workshop in Martigny, Switzerland, organized by European Association for Biometrics (EAB) in collaboration with the Center for Identification Technology Research (CITeR) and IDIAP research institute to present two progress report and one final project report.
- 2023.01: 🎉 Our journal on Multi-finger fingerprint has been publied in IET Biometrics journal.
- 2022.11 I received the best poster award in CITer Fall Program Review
- 2022.11 🔥 Awarded a research grant from CITer for our new project (Project #22F-01W), A Perpetual Deep Face Recognition System, which aims to build a dynamic deep learning model that continually learns new FR tasks.
- 2022.08: 🎉 One paper Multi-finger fusion is accepted in BIOSIG 2022
- 2022.04 🔥 Awarded a research grant from CITer for our new project (Project #22S-06W), One-to-One Face Recognition with Human Examiner in the Loop, which aims to improve the performance of a FR system with human examiner in the loop.
- 2022.01: 🔥 My code on installing CUDA in Ubuntu has exceeded 500 stars!(⭐️0.5k+)
- 2021.06: I joined the Deep Learning Lab at WVU to work under Prof. Nasser Nasrabadi as a PhD researcher!
- 2021.04: 🎉 Our journal on Gait Recognition has been publied in IET Computer Vision journal.
- 2020.09: I defended my M.Sc. thesis at IICT, BUET. Thanks to Dr. Hossen Asiful Mustafa for your invaluable supervision.
📝 Publications
🎼 Automatic Target Recognition
Contrastive Learning and Cycle Consistency-Based Transductive Transfer Learning for Target Annotation
Shoaib Meraj Sami, Md Mahedi Hasan, Nasser Nasrabadi, Raghuveer Rao
- We propose a hybrid contrastive learning base unpaired domain translation (H-CUT) network that achieves a significantly lower FID score. It incorporates both attention and entropy to emphasize the domain-specific region, a noisy feature mixup module to generate high variational synthetic negative patches, and a modulated noise contrastive estimation (MoNCE) loss to reweight all negative patches using optimal transport for better performance.
- Our proposed contrastive learning and cycle-consistency-based TTL (C3TTL) framework consists of two H-CUT networks and two classifiers.
🧑🎨 Face Recognition
Text-Guided Face Recognition using Multi-Granularity Cross-Modal Contrastive Learning
Md Mahedi Hasan, Shoaib Meraj Sami, and Nasser Nasrabadi. [code] [video]
- We introduce text-guided face recognition (TGFR) to analyze the impact of integrating facial attributes in the form of natural language descriptions while we hypothesize that adding semantic information into the loop can significantly improve the image understanding capability of an FR algorithm compared to other soft biometrics.
- We also design a face-caption alignment module (FCAM), which incorporates cross-modal contrastive losses across multiple granularities to maximize the mutual information between local and global features of the face-caption pair.
Improving Face Recognition from Caption Supervision with Multi-Granular Contextual Feature Aggregation
Md Mahedi Hasan, and Nasser Nasrabadi. [code]
- We introduce caption-guided face recognition (CGFR) as a new framework to improve the performance of commercial-off-the-shelf (COTS) face recognition (FR) systems.
- We propose a contextual feature aggregation module (CFAM) that addresses this issue by effectively exploiting the fine-grained word-region interaction and global image-caption association. Specifically, CFAM adopts a self-attention and a cross-attention scheme for improving the intra-modality and inter-modality relationship between the image and textual features, respectively.
- We also design a textual feature refinement module (TFRM) that refines the textual features of the pre-trained BERT encoder by updating the contextual embeddings.
📚 Fingerprint Recognition
On Improving Interoperability for Cross-Domain Multi-Finger Fingerprint Matching Using Coupled Adversarial Learning
Md Mahedi Hasan, Nasser Nasrabadi, and Jeremy Dawson
- We project both the contactless and the contact-based fingerprint into a latent subspace to explore the hidden relationship between them using class-specific contrastive loss and ArcFace loss.
- The ArcFace loss ensures intra-class compactness and inter-class separability, whereas the contrastive loss minimizes the distance between the subspaces for the same finger.
- Experiments on four challenging datasets demonstrate that our proposed model outperforms two top-performing commercial-off-the-shelf SDKs, i.e., Verifinger v12.0 and Innovatrics.
Deep Coupled GAN-Based Score-Level Fusion for Multi-Finger Contact to Contactless Fingerprint Matching [Oral Presentation]
Md Mahedi Hasan, Nasser Nasrabadi, and Jeremy Dawson.
- To improve the interoperability between contact to contactless images in fingerprint matching, we propose a coupled deep learning framework that consists of two Conditional Generative Adversarial Networks.
- Generative modeling is employed to find a projection that maximizes the pairwise correlation between these two domains in a common latent embedding subspace.
🎼 Gait Recognition
IET Computer Vision 2021Learning view-invariant features using stacked autoencoder for skeleton-based gait recognition, Md Mahedi Hasan, and Hossen Asiful Mustafa.IJCSIS 2021Multi-level feature fusion for robust pose-based gait recognition using RNN, Md Mahedi Hasan, and Hossen Asiful Mustafa.
Others
ETCCE 2021A deep Spatio-temporal network for vision-based sexual harassment detection, Md Shamimul Islam, Md Mahedi Hasan, Sohaib Abdullah, et al.Book Chapter 2020Deep Learning based Early Detection and Grading of Diabetic Retinopathy Using Retinal Fundus Images, Sheikh Muhammad Saiful Islam, Md Mahedi Hasan, Sohaib Abdullah arXivRAAICON 2019Robust Pose-Based Human Fall Detection using Recurrent Neural Network, Md Mahedi Hasan, Md Shamimul Islam, Sohaib AbdullahICBSLP 2019Aibangla: A Benchmark Dataset for Isolated Bangla Handwritten Basic and Compound Character Recognition, Md Mahedi Hasan, Mahathir Mohammad Abir, et al. [dataset]ICBSLP 2018YOLO-Based Three-Stage Network for Bangla License Plate Recognition in Dhaka Metropolitan City, Sohaib Abdullah, Md Mahedi Hasan, Sheikh Muhammad Saiful Islam [code], [dataset]ICCIT 2018DEEPGONET: Multi-label Prediction of GO Annotation for Protein from Sequence Using Cascaded Convolutional and Recurrent Network, Sheikh Muhammad Saiful Islam, Md Mahedi HasanIJCSIS 2020Forensic Detection of Digital Image Tampering Using Statistical Analysis, Md. Zahurul Haque, Md Mahedi Hasan
📝 Research Grants
Multi-Finger Contactless Fingerprint Matching
- PI Name: Jeremy M. Dawson, Nasser M. Nasrabadi
- Name of Funding Organization: CITer (Project #21S-04W), IUCRC - NSF
- Period of Grant Award: 1 Year (08/13/2021 - 08/12/2022)
- Amount: $50,000
- Project Title: Evaluation of the Performance of Multi-Finger Contactless Fingerprint Matching
-
My Role in the Project: I developed an algorithm for multi-finger contactless fingerprint matching. I successfully achieved all project milestones under the supervision of the PIs. I presented the final report and webinar, along with publishing two academic papers based on the experimental results.
One-to-One Face Recognition
- PI Name: Nasser M. Nasrabadi, Jeremy M. Dawson
- Name of Funding Organization: CITer (Project #22S-06W), IUCRC - NSF
- Period of Grant Award: 1 Year (11/05/2022 - 04/21/2023)
- Amount: $50,000
- Project Title: One-to-One Face Recognition with Human Examiner in the Loop
-
My Role in the Project: I developed a text-guided face recognition (FR) system to improve the performance of state-of-the-art FR algorithms by integrating facial attributes through natural language descriptions. I successfully met all project milestones under the supervision of the PIs. I presented both the progress report and the final report, and additionally published two academic papers.
Deep Face Recognition
- PI Name: Nasser M. Nasrabadi, Md Mahedi Hasan
- Name of Funding Organization: CITer (Project #22F-01W), IUCRC - NSF
- Period of Grant Award: 1 Year (09/03/2022 - 10/25/2023)
- Amount: $50,000
- Project Title: A Perpetual Deep Face Recognition System
-
My Role in the Project: I wrote the proposal with Prof. Nasser. I designed the class-incremental learning framework which can learn and improve from a sequence of face recognition tasks without storing any exemplar sets. I successfully completed all the project milestones. I presented the progress and the final report.
📝 Research Projects
Medical Image Classification
- Project Title: Early Detection and Grading of Diabetic Retinopathy Using Retinal Fundus Images
- Period: 1 Year (10/14/2017 - 10/30/2018)
- Description: We developed a novel deep convolutional neural network, which performs the early-stage detection by identifying all microaneurysms (MAs), the first signs of DR, along with correctly assigning labels to retinal fundus images which are graded into five categories. We have tested our network on the largest publicly available Kaggle diabetic retinopathy dataset, and achieved 0.851 quadratic weighted kappa score and 0.844 AUC score.
- Resources: [paper]
License Plate Recognition
- Project Title: Real-time Automatic Bangla License Plate Detection and Recognition
- Period: 1 Year (05/01/2018 - 04/30/2019)
- Description: We have developed a real-time automatic Bangla license plate recognition system based on YOLO-v3. Additionally, we curated a dataset comprising 1,500 diverse images of Bangladeshi vehicular license plates. These images were manually captured from streets, simulating various real-world scenarios. This project is funded by department of CSE, Manarat International University (MIU).
- Resources: [dataset], [paper]
Bangla Handwritten Character Recognition
- Project Title: Deep Isolated Bangla Handwritten Basic and Compound Character Recognition
- Period: 1 Year (01/12/2018 - 12/30/2018)
- Description: We present AIBangla, a new benchmark image database for isolated handwritten Bangla characters with detailed usage and a performance baseline. Our dataset contains 80,403 hand-written images on 50 Bangla basic characters and 249,911 hand-written images on 171 Bangla compound characters which were written by more than 2,000 unique writers from various institutes across Bangladesh. This project is funded by department of CSE, Manarat International University (MIU)
- Resources: [dataset], [paper]
Bangla Sign Language Recognition
- Project Title: Deep Bangla Sign Language Recognition from Video: A New Large-scale Dataset
- Period: 2 Year (11/05/2021 - 04/05/2023)
- Description: We have developed an attention-based Bi-GRU model that captures the temporal dynamics of pose information used by individuals communicating through sign language. Furthermore, we created a large-scale dataset called the MVBSL-W50, which comprises 50 isolated words across 13 categories.
- Resources: [dataset], [paper]
🎖 Awards
- 2022.11 Best Poster Award in CITer Fall Program Review
- 2009-2012 University Merit Scholarship (Undergraduate) (All eight terms of undergraduate study from the Govt. of the People’s Republic of Bangladesh)
📖 Educations
2021.16 -
- West Virginia University (WVU)
- Doctor of Philosophy degree in computer engineering (CE)
- Advisor: Dr. Nasser M. Nasrabadi, Professor, LCSEE, WVU
2014.10 - 2020.09
- Bangladesh University of Engineering and Technology (BUET)
- M.Sc. in Information and Communication Technology (ICT)
- Thesis: View Invariant Gait Recognition for Person Re-Identification in a Multi Surveillance Camera Environment
- Advisor: Dr. Hossen Asiful Mustafa, Professor, IICT, BUET
2008.12 - 2013.09
- Khulna University of Engineering and Technology (KUET)
- B.Sc. in Electrical and Electronic Engineering
- Advisor: Dr. Mohiuddin Ahmad, Professor, EEE, KUET
2006.10 - 2008.04
- Higher Secondary Certificate (H.Sc.)
- Engineering University Higher Secondary School
💬 Academic Activities
- Graduate researcher at West Virginia University (WVU)
- Graduate researcher at Center for Identification Technology Research (CITer)
- Reviewer at IEEE Access
- Reviewer at PLOS ONE
- Reviewr at Engineering Applications of Artificial Intelligence, ELSEVIER
💬 Teaching
2019.04 - 2021.05
- Teaching Faculty at Manarat International University (MIU), Department of Computer Science and Engineering (CSE)
- Responsibilities: Conducting undergraduate classes
- CSE-437: Computer Vision and Robotics [Spring 2019], [Fall 2019]
- CSE-411: Artificial Intelligence [Summer 2019]
- CSE-433: Neural Networks and Fuzzy Systems [Fall 2019]
- Supervising undergraduate student research and projects
💬 Affiliations
2015.03 - 2021.05
- Assistant Editor of Byapon Science Magazine
- Bi-monthly youth science magazine, Printed and circulated nationwide more than 15,000 copies per issue
- Office: 48/1, Motijheel C/A, Dhaka-1200, Bangladesh
2023.07 - Present
- IEEE Student Member
💬 Invited Talks
- 2023.12, The Engine Behind ChatGPT. Cognitive Mind Webinar, Dhaka, Bangladesh. [video]
- 2019.11, Deep Learning in Computer Vision. Google Developer Group, Sonargaon, Dhaka, Bangladesh. [video]
- 2020.03, Gait Recognition. CSE Fest 2020, MIU, Dhaka, Bangladesh (March 2020)