Research Assistant · Research Engineer · Multimodal & Trustworthy AI

Building explainable, data-driven AI for medical imaging, multimodal perception, and decision support.

I develop vision transformer architectures, multimodal fusion pipelines, and explainable AI systems across medical imaging, NLP, and deployment-oriented applications—spanning clinical classification, visual question answering, and real-time web tools.

Medical Imaging Vision Transformers Multimodal Fusion Explainable AI NLP IoT / Threat Detection
Open Scholar LinkedIn Email Open CV
Scroll down ↓
Rezaul Haque

Quick highlights

Peer-reviewed journals: 14 · Under review: 8 · Selected conference papers: 12 · Best Paper awards: 6

  • Research Assistant at East West University (CRT) (Jan 2024–Present).
  • Research Engineer at Dataxense (AI Division) (Feb 2023–Jun 2025).
  • Trained 300+ students in medical image analysis, trustworthy AI, and research workflows.

Contact: rezaulh603@gmail.com · +8801314996147

Experience Publications

Featured projects

Selected applied work and collaborations from my CV.

View all

Multimodal Glaucoma Detection (EWU CRT × Vision Eye Hospital)

Nov 2025

MedicalMultimodalSemi-supervised
  • Semi-supervised self-attention model for calibrated multimodal fusion to support real-time clinical screening.
Scholar Details

Industrial Fabric Defect Detection (EWU CRT × Knitasia Ltd.)

Aug 2024

IndustryReal-time
  • Retrofitted CV system for defect detection and grading with a web dashboard for factory deployment.
Scholar Details

Multi-stage Leukemia Classification Pipeline

Feb 2023 – Jun 2025

MedicalTransfer learning
  • Built a multi-stage pipeline using Inception-ResNetv2, optimized sensitivity/precision for rare subtypes and CDS integration.
Scholar Details

Underwater Object Detection (Cross-stage multi-branch + Large-kernel SPP)

Feb 2023 – Jun 2025

DetectionCV
  • Designed a novel detector with multi-branch fusion and large-kernel spatial pyramid modules to improve mAP and speed on diverse underwater datasets.
Scholar Details

Selected publications

Buttons open Google Scholar search for each title.

Browse all

Vision-Audio Multimodal Object Recognition Using Hybrid and Tensor Fusion Techniques

Information Fusion (Q1), 2025

JournalMultimodal
Scholar

Explainable deep stacking ensemble model for accurate and transparent brain tumor diagnosis

Computers in Biology and Medicine (Q1), 2025

JournalXAI
Scholar

DepTformer-XAI-SV: A Novel Ensemble Transformer Model for Fast and Accurate Depression Emotion and Severity Analysis

iScience (accepted)

AcceptedNLP
Scholar

LightVTD: Lightweight Explainable Vision Transformer with Multi-Path Token Fusion for Drowsiness Detection

Scientific Reports (accepted)

AcceptedXAI
Scholar