Research

Publications, ongoing work and research interests in machine learning, computer vision and intelligent healthcare.

01Philosophy

Research with practical impact.

My research focuses on developing machine learning systems that solve practical real-world problems while maintaining scientific rigor and reproducibility. I am particularly interested in computer vision, image processing, intelligent healthcare and document understanding, with the long-term goal of contributing to impactful academic research.

02Publications

Featured publications.

Published2025

ChakmaCNN: A Pronunciation-Aware CNN for Handwritten Chakma Script Recognition

Md. Ragib Hasan, et al.

IEEE ConferenceFirst Author

A pronunciation-aware convolutional neural network for handwritten Chakma script recognition, designed to capture script-specific visual and phonetic patterns.

Read Paper

Published2025

DateFNet: From Pixels to Plates, Attention Based Multi-Stream CNN for Date Fruit Classification

Md. Ragib Hasan, et al.

IEEE ConferenceThird Author

An attention-based multi-stream CNN framework for classifying date fruits from image data, combining spatial feature extraction with attention mechanisms.

Read Paper

Published2026

Towards Secure Digital Communication: Deep Learning-Based Automated Classification of Malicious Bangla Messages

Md. Ragib Hasan, et al.

IEEE ConferenceFifth Author

A deep learning approach for automatically detecting and classifying malicious Bangla text messages to support safer digital communication.

Read Paper

03Review

Under review and submissions.

Performance and Generalization Analysis of CNN and Hybrid Deep Learning Models for Handwritten Medicine-Name Recognition in Bangladeshi Prescriptions

Intelligence-Based Medicine, Elsevier · Jul 2026First Author

RGCF-YOLO11: Region-Gated Context Fusion for Tiny PCB Defect Detection on Periodic Textures

ICCIT 2026 Conference SubmissionFirst Author

Content-Adaptive Multi-Scale Tokenization for Lightweight ViTs in Remote Sensing

ICCIT 2026 Conference SubmissionFirst Author

BongoBarta: A GRU-Attention Model with FastText Embeddings for Bengali News Headline Classification

ICCIT 2026 Conference SubmissionSecond Author

YGP-Net: A Lightweight P6-Enhanced YOLO11 Network for Efficient Obstacle Detection

ICCIT 2026 Conference SubmissionSecond Author

Two additional ICCIT 2026 conference submissions

ICCIT 2026 Conference SubmissionThird Author

04Thesis

Thesis and ongoing research.

From Traffic Scenes to Registration Strings: A Field Study of Vehicle Detection and Bangla License Plate Recognition in Bangladesh

Completed Thesis
First Author
YOLO
OCR
Bangla LPR

Supervisor: Md. Abu Naser Mojumder, Associate Professor and Head, Department of CSE, Sylhet Engineering College. • Collected and curated real-world traffic video data from five locations across Dhaka and Sylhet, building a field-derived dataset for vehicle detection and Bangla license plate recognition under unconstrained conditions. • Investigated small-object localization, degraded plate imagery, and complete Bangla registration-string recognition through a systematic evaluation of 13 detection and 9 OCR configurations; YOLO11n-SOEN achieved 0.6142 mAP@0.50:0.95, while BPNet achieved 82.36% test accuracy and 67.07% unseen-string accuracy. • Developed an end-to-end video inference pipeline integrating vehicle detection and plate recognition. I plan to continue further work on this topic.

Lightweight CNN Models for Handwritten Hangul Alphabet Classification

Research in Progress
CNN
Handwritten Script Recognition

Investigating lightweight handwritten-script recognition with CNN-based models for efficient recognition in low-resource settings.

05Interests

Research interests.

Computer Vision
Machine Learning
Deep Learning
Image Processing
Medical AI
OCR
Object Detection
Natural Language Processing
Pattern Recognition

Interested in collaborating on research?

Contact Me