Hi, I'am The

AI Researcher

As a postdoctoral researcher at IVCLab, Sejong University, since September 2025, my work centers on developing and applying AI methodologies to solve challenges in image processing and computer vision.In parallel, I independently develop LLM-based projects, exploring integration with VLMs.

Contact Me

Scroll Down

About Me

My Introduction

I obtained my Ph.D. in Intelligent Mechatronics Engineering from Sejong University, Seoul, South Korea, in August 2025. Prior to my doctoral studies, I worked as an Innovation Engineer at Samsung Electronics HCMC CE Complex in Ho Chi Minh City, Vietnam. I earned my Bachelor's degree in Mechatronics Engineering from Ho Chi Minh City University of Technology in 2019 through the PFIEV program, which is equivalent to the European Master 1.

06+ ⏳Years
experience

11+ 📚International
publications

02+ 🏢Organizations
worked

Download CV

Qualification

My Personal Journey

Education

Work

PhD in Intelligent Mechatronics Engineering

Sejong Univerisy, South Korea

Sep 2020 - Aug 2025

Bachelor of Mechatronics Engineering

HCM City University of Technology, Vietnam

Sep 2014 - Jul 2019

Postdoctoral Researcher

Intelligent Visual Computing Lab, Sejong University, South Korea

Sep 2025 - Present

Lead Researcher

Intelligent Visual Computing Lab, Sejong University, South Korea

Sep 2020 - Aug 2025

AI research Intern

PIAspace Co., Ltd, South Korea

Jun 2024 - Aug 2024

System Innovation Engineer

Samsung Electronics HCMC CE Complex (SEHC), Vietnam

Sep 2019 - Aug 2020

Skills

My Technical Level

Frameworks & Libraries

Pytorch Tensorflow OpenCV Ultralytics Ollama Langchain

Computer Vision & Image Processing

Image Restoration Object Detection Medical Imaging

Natural Language Processing

Video Captioning Prompting LLMs RAG

Deep Learning

Model Design CNNs Transformer (Vit, Swin) LSTM

Python

6 years

C/C++

3 years

C#

1 years

MATLAB

4 years

Javascript

2 years

Frontend

ReactJS Svelte Next.js TailwindCSS

Backend

Node.js FastAPI Celery SQL/NoSQL Docker

Vietnamese

Native

English

IELTS 6.5

French

DELF B1

Publications

My Research Contributions

T. V. Le, et al., “Surface Defect Detection Using Distributed Features”, Engineering Applications of Artificial Intelligence, 2025.
T. V. Le, J. Y. Lee, “QformerID: Quaternion Transformer Based Image Denoising”, IEEE Multimedia, 2025.
A. Telili, W. Hamidouche, I. Farhat, H. Amirpour, C. Timmerer, I. Khadraoui, J. Lu, T. V. Le, J. Baek, J. Y. Lee, Y. Wei, X. Sun, Y. Gao, J. Huang, Y. Zhong, “360-degree Video Super Resolution and Quality Enhancement Challenge: Methods and Results”, Signal Processing: Image Communication, 2025.
T. V. Le, J. Y. Lee, “Specular Highlight Removal Using Quaternion Transformer”, Computer Vision and Image Understanding, vol. 249, 104179, pp. 1–11, Dec. 2024.
Y. Choi, T. V. Le, G. Bang, J. Lee, J. Y. Lee, “Efficient Immersive Video Coding Using Specular Detection for High Rendering Quality”, Multimedia Tools and Applications, 2024.
T. V. Le, J. Y. Lee, “Quality Enhancement Based Video Captioning in Video Communication Systems”, IEEE Access, 2024.
Y. B. Kim, T. V. Le, J. Y. Lee, “Lightweight Brain MR Image Super-Resolution Using 3D Convolution”, Multimedia Tools and Applications, vol. 83, no. 3, pp. 8785–8795, Jan. 2024.
J. Y. Lee, Y. Choi, T. V. Le, K. Choi, “Efficient Feature Coding Based on Performance Analysis of Versatile Video Coding (VVC) in Video Coding for Machines (VCM)”, Multimedia Tools and Applications, vol. 82, no. 27, pp. 42803–42816, Nov. 2023.
T. V. Le, Y. Choi, J. Y. Lee, “Specular Detection and Rendering for Immersive Multimedia”, IEEE Multimedia, vol. 30, no. 2, pp. 38–47, Apr.–Jun. 2023.
J. Y. Lee, T. V. Le, Y. Choi, K. Choi, “Low-Complexity Two-Step Lossless Depth Coding Using Coarse Lossy Coding”, Multimedia Tools and Applications, vol. 81, no. 10, pp. 14065–14079, Apr. 2022.
K. Choi, T. V. Le, Y. Choi, J. Y. Lee, “Low-Complexity Intra Coding in Versatile Video Coding”, IEEE Trans. Consumer Electronics, vol. 68, no. 2, pp. 119–126, May 2022.

CV Projects

Most Recent Works

Dissertation: Deep learning based specular highlight detection and removal (SHDRN)

💡 Proposes a two-stage SHDRN framework for accurate specular highlight detection and removal in natural images.
🎯 Integrates VAUnet for precise specular mask generation and QformerSHR for high-quality removal.
🌀 QformerSHR leverages Quaternion representation with DSQConv and QTB for efficient global context learning.
🚀 Outperforms state-of-the-art CNN-based methods in visual quality and quantitative metrics.

Tech stack: PyTorch, Python, Transformer, Quaternion Networks, Image Restoration

Surface Defect Detection Using Distributed Features (DF-SDD)

🧠 Proposes a novel low-complexity surface defect detection network (DF-SDD).
🧩 Introduces Distributed Convolution (DConv) and Distributed Depthwise Convolution (DDWConv).
🌐 Combines multi-scale features with Spatial Attention Gate (SAG) for enhanced fusion.
📈 Outperforms state-of-the-art networks (YOLOv8, Faster R-CNN, etc.) in mAP and complexity.

Tech stack: PyTorch, Python, Object Detection, Network Design

Specular Detection and Rendering for Immersive Multimedia

🧠 Proposes VAUnet: a novel deep learning model with VAE and attention for specular highlight detection.
🧩 Introduces a hybrid loss (dice + cross-entropy + KL divergence) for better mask accuracy.
🎯 Achieves superior accuracy specular detection compared to existing methods (SHDnet, JSHDR, etc.).
📦 Applies detection masks to MPEG-I rendering to enhance realism in immersive video.

Tech stack: PyTorch, Python, MPEG-I, C++, Segmentation, Network Design

Lightweight 3D CNN for Brain MRI Super-Resolution

🧠 Proposes a lightweight 3D convolutional network for high-quality brain MRI super-resolution.
⚡ Designed with extraction, compression, and reconstruction modules for efficiency.
📉 Reduces network parameters for real-time feasibility in MR scanners.
📊 Outperforms conventional networks in PSNR/SSIM while lowering complexity.

Tech stack: PyTorch, Python, 3D CNN, Medical Imaging, MRI Processing

LLM Projects

Most Recent Works

AI Interview: Transforming the Future of Recruitment

📄 Automates question generation based on job descriptions (JDs) using large language models (LLMs).
🗣️ Integrates large language models (LLMs) to conducts natural voice interviews with dynamic Q&A.
🌐 Web UI for data management,job description input,managing data, and real-time interviewing.
📊 Provides unbiased, consistent evaluations and candidate ranking based on multiple criteria.
🗃️ Stores candidate data, transcriptions, and reports in an integrated system.

Tech stack: Svelte, TailwindCSS (frontend); FastAPI (backend)

Demo

CV Screening: Automating Candidate Evaluation

📄 Extracts and standardizes information from diverse CV formats using LLMs and VLMs.
🛠️ Utilizes models such as DeepSeek, Owen, OpenAI API, LlamaIndex, and OllamaOCR for accurate data extraction.
📊 Scores and ranks candidates based on customizable criteria including education, skills, experience, and roles.
🌐 Features a web interface for uploading CVs and job descriptions, managing data, and viewing evaluation results.
⚙️ Stores candidate data in database to simplify management and integration.

Tech stack: Svelte, TailwindCSS (frontend); FastAPI (backend)

Demo

Multimodal RAG: Intelligent Document Understanding

📄 Automatically extracts and summarizes text, tables, and images from complex documents, retaining only content-relevant images using VLMs and LLMs.
🗣️ Uses LLMs to answer user queries with context-aware reasoning.
🔍 Leverages MultiVectorRetriever to efficiently search across multiple embeddings (text, tables, and image captions) for precise retrieval.
🌐 Provides a Web UI for uploading PDFs, viewing extraction results, and interacting via a chat interface.

Tech stack: Svelte, TailwindCSS (frontend); FastAPI, LangChain, Ollama (backend)

Demo