Most Nilufa Yeasmin

PhD Student · Dept. of Electrical Engineering & Computer Science, South Dakota State University

Most Nilufa Yeasmin

PhD Student in Computer Science (AI & Computer Vision) · Advised by Dr. Ruyi Lian, South Dakota State University

Portrait of Most Nilufa Yeasmin

I study computer vision and 3D geometric reasoning, with a focus on making camera pose estimation robust when correspondences are noisy or wrong. My current work centers on DeepOnP, a neural solver for the Orthographic-n-Point problem that combines top-K rotation hypotheses with quaternion averaging to stay reliable under high outlier ratios and image noise — broadly relevant to 3D reconstruction, multi-view geometry, and any pipeline that depends on trustworthy pose estimates.

I am a PhD student in the Department of Electrical Engineering and Computer Science at South Dakota State University, advised by Dr. Ruyi Lian. Before starting my PhD, I was a research assistant at the BUBT Research and Innovation Center in Dhaka, Bangladesh, and completed my B.Sc. in Information and Communication Technology at Islamic University, Bangladesh, where my earlier work spanned phishing website detection, biomedical network analysis, and COVID-19 sentiment analysis.

Location
Brookings, SD, USA
Email
MostNilufa.Yeasmin [at] sdstate [dot] edu
Phone
+1 605-691-0200

At a glance

Google Scholar →
8
Publications
242
Citations
3
First-author journal papers
2
Research labs / institutions

Citation count from Google Scholar. h-index / i10-index not shown yet — update once available.

Research Interests

01

Computer Vision

Robust perception pipelines for real-world, noisy visual data.

02

3D Reconstruction & Multi-view Geometry

Recovering geometry and structure from images and correspondences.

03

Camera Pose Estimation

Outlier-resilient solvers for the Orthographic-n-Point problem.

04

Geometric Deep Learning & Neural Solvers

Learned solvers that combine classical geometry with neural priors.

05

Foundation Models for Vision

Adapting large pretrained vision models to geometric estimation tasks.

06

Medical Image Analysis & Explainable ML

Segmentation, diagnosis support, and interpretable model behavior.

News

Selected Publications

All publications →

Biography

Most Nilufa Yeasmin is a PhD student in Computer Science at South Dakota State University, working under the supervision of Dr. Ruyi Lian on robust, geometry-driven computer vision. Her dissertation research, DeepOnP, targets outlier-resilient neural solvers for camera pose estimation — a problem central to 3D reconstruction and multi-view geometry.

Before her PhD, she worked as a research assistant at the BUBT Research and Innovation Center in Dhaka, and earlier, as an undergraduate at Islamic University, Bangladesh, where she built machine learning pipelines for COVID-19 Twitter sentiment analysis and phishing website detection — work that led to several peer-reviewed journal publications. She also contributed to biomedical research on network-based biomarker identification and high-throughput biological sequence analysis at the Computer Vision and Intelligent Interfacing Lab.

She has served as an ad hoc reviewer for PLOS ONE, mentored students in physics and mathematics, and remains an active member of student research and community organizations, including Alorito'30 and the Chapai Nawabganj Student Association in Bangladesh.