Computer Vision

Our research group aims to develop translational research and products that enhance millions of people’s lives. Noticing immeasurable real-life problems relating to image, video, and sensory data, we push advancing research on computer vision. Humans are at the crux of our research, epitomized by a wide range of research topics such as face recognition and manipulation, eye gaze prediction, hand gesture recognition, and human behavior understanding. Another complementary strand is to understand the formulation of real-world imagery data, rebuild, and manipulate them, realized by Generative AI research.

To make computer vision algorithms work in real-life scenarios, we identify practical challenges, including data scarcity and data quality degradation, and resolve them via advanced technologies in Few-shot learning and Image/Video Enhancement. We do not limit our research to imagery data but extend our research to cover other sensory data, such as 3D point-cloud, as well as combining vision with other modalities like languages. Our computer vision research, therefore, supplies impactful research and products to enhance human life such as smart mobility and smart surveillance systems, deployed on thousands of smart cars and smart cameras in Vietnam.
The Computer Vision team has helped boost the global visibility of VinAI by establishing a strong collaborator network with prominent researchers all over the world. We achieved substantial research outputs at top-tier AI venues, under a wide range of, but not limited to, the following topics:
- Face recognition and analyses
- Human activity understanding
- Image generation and manipulation
- Few-shot learning
- Image/Video enhancement
- 3D Vision
- Vision and language
- Trustworthy Computer Vision

CV CVPR Top Tier
Clustering Plotted Data by Image Segmentation

Clustering is a popular approach to detecting patterns in unlabeled data. Existing clustering…

CV UAI Top Tier
Simple Transferability Estimation for Regression Tasks

Transfer learning has been a widely used technique to adapt a deep learning…

CV ICCV Top Tier
GaPro: Box-Supervised 3D Point Cloud Instance Segmentation Using Gaussian Processes as Pseudo Labelers

Instance segmentation on 3D point clouds (3DIS) is a longstanding challenge in computer…

CV ICCV Top Tier
Conditional 360-degree Image Synthesis for Immersive Indoor Scene Decoration

In this paper, we address the problem of conditional scene decoration for 360-degree…

Related publications

CV CVPR Top Tier
March 6, 2024

Supreeth Narasimhaswamy, Huy Nguyen, Lihan Huang, Minh Hoai

CV CVPR Top Tier
March 6, 2024

Ka Chun Shum, Jaeyeon Kim, Binh-Son Hua, Duc Thanh Nguyen, Sai-Kit Yeung

CV CVPR Top Tier
March 6, 2024

Phong Tran, Egor Zakharov, Long-Nhat Ho, Anh Tran, Liwen Hu, Hao Li

CV CVPR Top Tier
March 6, 2024

Trung Tuan Dao, Duc Hong Vu, Cuong Pham, Anh Tran

CV CVPR Top Tier
March 4, 2024

Bang-Dang Pham, Phong Tran, Anh Tran, Cuong Pham, Rang Nguyen, Minh Hoai

Do not miss these Seminars & Workshops

Huu Le

Chalmers University of Technology

Robust Parameter Estimation in Computer Vision
Wed, Jan 8 2020 - 03:00 pm (GMT + 7)

Stefano Ermon

Stanford University

Learning with Limited Supervision
Fri, Aug 16 2019 - 10:00 am (GMT + 7)

Duc Nguyen

Yonsei University

Deep Learning for Analysis and Reconstruction of 3D Shapes as Point Clouds and Meshes
Fri, Mar 10 2023 - 02:30 pm (GMT + 7)

Released Source Codes








Anti-DreamBooth: Protecting users from personalized text-to-image synthesis ICCV 2023


BERTweet: A pre-trained language model for English Tweets EMNLP 2020


BARTpho: Pre-trained Sequence-to-Sequence Models for Vietnamese InterSpeech 2021

Technical Blog