Nhat-Tan Bui

Staff Research Associate, Robotics Institute, Carnegie Mellon University

"True scholar is shown by phronesis, not imagination."

Email: tannhatb@andrew.cmu.edu

Greetings! My name in Vietnamese is Bùi Nhật Tân (裴日新), which typically means daily renewal 🌕 with brightness 🔆 and hope 🍀. I'm affiliated with the Human Sensing Lab, working with Fernando De la Torre, who is truly a mensch and a wonderful advisor!

I was fortunate to receive my M.S. from the University of Arkansas (UARK 🇺🇸), advised by Susan Gauch and Truong Nguyen (UCSD), and my B.S. from the University of Science, VNU-HCM (HCMUS 🇻🇳) and Auckland University of Technology (AUT 🇳🇿), advised by Minh-Triet Tran and Ngoc-Thao Nguyen. I graduated from Le Quy Don High School, the oldest school in Ho Chi Minh City (Saigon).

Research Keywords: Deep Learning 🧠, Computer Vision 👀, and Multimodal Learning 🎏 with applications in 3D/Video Understanding 🎞️, Spatial Understanding 🛰️, and Vision-Language Actions 🤖.

I'm looking for Research Engineer/PhD positions starting in 2027. Please reach out if you feel I'm a good fit!!!

Publications

(* indicates equal contribution)

CoVeR: Coverage-Based Token Pruning for Multi-View 3D Reasoning in VLMs

Under Review . More details coming soon 👾

Nhat-Tan Bui*, Varshini Elangovan*, Arun Reddy Anugu, Sreyas Mohan, Wei Ye, Dilin Wang, JQ Huang, Rakesh Ranjan, Aviral Chharia, Fernando De la Torre

A training-free, geometry-only token pruning method for multi-view 3D VLMs that maximizes scene coverage.

[Website]

NeIn: Telling What You Don't Want

Conference on Computer Vision and Pattern Recognition Workshops (CVPRW 2025, 🏆 Best Paper Award)

Nhat-Tan Bui, Dinh-Hieu Hoang, Quoc-Huy Trinh, Minh-Triet Tran, Truong Nguyen, Susan Gauch

A large-scale benchmark for teaching image editing models to understand negation.

[Paper] [Website] [Master's Thesis]

TSRNet: Simple Framework for Real-time ECG Anomaly Detection with Multimodal Time and Spectrogram Restoration Network

International Symposium on Biomedical Imaging (ISBI 2024)

Nhat-Tan Bui*, Dinh-Hieu Hoang*, Thinh Phan, Minh-Triet Tran, Brijesh Patel, Donald Adjeroh, Ngan Le

A real-time framework that spots abnormal heartbeats in ECG signals by fusing time and spectrogram.

[Paper] [Code]

SAM3D: Segment Anything Model in Volumetric Medical Images

International Symposium on Biomedical Imaging (ISBI 2024)

Nhat-Tan Bui*, Dinh-Hieu Hoang*, Minh-Triet Tran, Gianfranco Doretto, Donald Adjeroh, Brijesh Patel, Arabinda Choudhary, Ngan Le

A lightweight framework that turns the SAM into a 3D segmenter for volumetric medical scans.

[Paper] [Code]

MEGANet: Multi-Scale Edge-Guided Attention Network for Weak Boundary Polyp Segmentation

Winter Conference on Applications of Computer Vision (WACV 2024)

Nhat-Tan Bui, Dinh-Hieu Hoang, Quang-Thuc Nguyen, Minh-Triet Tran, Ngan Le

A segmentation network that recovers weak polyp boundaries by using Laplacian edge information to guide attention.

[Paper] [Code]

Documents

Credits/Misc