Khang Truong Giang

PhD candidate at NMAIL, KAIST.

prof_pic.jpeg

Room 3552, E3-2 building

KAIST CS, Daejeon 34141, Republic of Korea

Hello, my name is Khang Truong Giang, a final-year PhD candidate in Computer Science at KAIST. I am fortunate to be advised by Prof. Sungho Jo and co-advised by Prof. Soohwan Song. Before joining KAIST, I did my Bachelor at Hanoi University of Science and Technology (HUST), Vietnam. During this time, I worked on some projects related to topic modeling, machine learning.

Research interests: Currently, I have focused on addressing challenging problems in 3D reconstruction, which involves building a 3D model of a scene using monocular images. I have developed various advanced deep-learning models tailored to resolve several tasks in 3D reconstruction, including feature matching, visual localization, and multi- view depth estimation. Furthermore, I actively contribute research codes on Github, which are widely utilized by hundreds of users globally.

I am looking for a new role. I am grateful for any advices and opportunities!

news

Feb 26, 2024 Our visual localization paper is accepted at CVPR 2024.
Nov 20, 2022 Two paper accepted at AAAI-23 and Pattern Recognition! :smile:

selected publications

  1. Learning to Produce Semi-dense Correspondences for Visual Localization (Oral, top 3.3% (90/2719) of accepted papers)
    Khang Truong Giang, Soohwan Song^, and Sungho Jo^
    In Computer Vision and Pattern Recognition Conference (CVPR) 2024
    PDF Code
  2. TopicFM: Robust and Interpretable Topic-assisted Feature Matching
    Khang Truong Giang, Soohwan Song^, and Sungho Jo^
    In Proceedings of the AAAI conference on artificial intelligence 2023
    PDF Code
  3. Prior Depth-Based Multi-View Stereo Network for Online 3D Model Reconstruction
    Soohwan Song*, Khang Truong Giang*, Daekyum Kim, and Sungho Jo
    Pattern Recognition 2023
    PDF Code
  4. CURVATURE-GUIDED DYNAMIC SCALE NETWORKS FOR MULTI-VIEW STEREO
    Khang Truong Giang, Soohwan Song, and Sungho Jo
    In International Conference on Learning Representations 2022
    PDF Code
  5. Sequential Depth Completion With Confidence Estimation for 3D Model Reconstruction
    Khang Truong Giang*, Soohwan Song*, Daekyum Kim, and Sunghee Choi
    IEEE Robotics and Automation Letters 2020
    PDF Code