Sithu Aung

I was a M.S student at the Visual Intelligence Group, part of the Korea Institute of Science and Technology at University of Science and Technology, Korea, where I work on 3D computer vision. My M.S advisor is Junghyun Cho.

Email  /  GitHub  /  Google Scholar

Research

I'm interested in 3D Computer Vision, Scene Reconstruction & Understanding, Digital Twins.

Pseudo-Labeling via Video Object Tracking for Scalable Monocular 3D Object Detection

A training-free pseudo-labeling framework for 3D object detection.

Multi-View Pedestrian Occupancy Prediction with a Novel Synthetic Dataset

A new occupancy prediction dataset for dense pedestrians in multi-view environments.

Enhancing Multi-view Pedestrian Detection Through Generalized 3D Feature Pulling

An effective 3D feature pulling method for generalizable multi-view pedestrian detection.

Open-source Projects

I open-sourced a lot of works in my earlier days.

Easy-to-use Face Analysis Tool

Face analysis tool including face & landmark detection, recognition, facial expression recognition and facial attribute classification.

A collection of SOTA Image Classification Models in PyTorch

Intended for easy to use and integrate SOTA image classification models into down-stream tasks and finetuning with custom datasets.

SOTA Segmentation Models in PyTorch

Easy to use and customizable SOTA semantic segmentation models in PyTorch.

Top-Down Multi-person Pose Estimation

Multi-person pose estimation based on top-down method, incorporating real-time YOLO detector with SOTA pose estimation models.

Object Tracking with YOLO, CLIP and DeepSORT

Tracking by detection with real-time YOLO detector with classic DeepSORT tracking pipeline and zero-shot feature extractor, CLIP.

Services

Reviewer: ICCV 2025, BMVC 2025, CVPR 2026.

Design and source code from Jon Barron's website