About Me

Hi ~ I’m Le-Anh Tran, PhD, an AI Researcher/Engineer fascinated by neural networks - what makes them so cool, how they do surprisingly clever things, and why they are occasionally stupid. I enjoy turning ideas into experiments and making AI not only smarter, but genuinely useful in the real world.

My work focuses on Image Enhancement & Understanding, where I spend most of my time teaching machines to see a little better (though I explore other areas as well). I enjoy the full journey from a questionable idea to an algorithm that actually works. I read papers, write papers, and occasionally discover that someone else had already had the exact same idea three years earlier.

I'm also a big fan of 3-cushion billiards and soccer. I spend my free time trying to convince balls to go where I want them to.



Email: tranlevision@gmail.com
Profiles: ResearchGate | GoogleScholar | MyResume
Projects | Blogs: Github | Medium


Education

PhD in Computer Vision

3/2021 – 2/2024
Myongji University, South Korea

MSc in Computer Vision

3/2019 – 2/2021
Myongji University, South Korea

BEng in Automation and Control Engineering

9/2014 – 8/2018
HCMC University of Technology and Engineering (HCMUTE), Vietnam


Experience

AI Engineer

1/2026 – Now
DeltaX Co., Ltd., Seoul, South Korea

AI Researcher

4/2020 – 5/2025
MindinTech, Inc., Seoul, South Korea

Software Developer (Intern)

7/2019 – 9/2019
OCST Co., Ltd., Seoul, South Korea

AI Engineer

3/2018 – 2/2019
FPT Software, Ho Chi Minh City, Vietnam

Teaching Assistant

2/2017 – 1/2018
Faculty of Electrical and Electronics Engineering, HCMUTE, Vietnam




Publications

Unpaired Image Dehazing via Kolmogorov-Arnold Transformation of Latent Features

LA Tran
Pattern Recognition, 2026

Distilled Pooling Transformer Encoder for Efficient Realistic Image Dehazing

LA Tran, DC Park
Neural Computing and Applications, 2025

Lightweight Image Dehazing Networks based on Soft Knowledge Distillation

LA Tran, DC Park
The Visual Computer, 2024

Encoder-Decoder Networks with Guided Transmission Map for Effective Image Dehazing

LA Tran, DC Park
The Visual Computer, 2024

Cluster Analysis via Projection onto Convex Sets

LA Tran, D Kwon, HM Deberneh, DC Park
Intelligent Data Analysis, 2024

Toward Improving Robustness of Object Detectors Against Domain Shift

LA Tran, NC Tran, DC Park, J Carrabina, D Castells-Rufas
GECOST 2024, IEEE

Single Image Dehazing via Regional Saturation-Value Translation

LA Tran, D Kwon, DC Park
Procedia Computer Science, 2024

POCS-based Clustering Algorithm

LA Tran, HM Deberneh, TD Do, TD Nguyen, MH Le, DC Park
IWIS 2022, IEEE

A Novel Encoder-Decoder Network with Guided Transmission Map for Image Dehazing

LA Tran, S Moon, DC Park
Procedia Computer Science, 2022

Robust U-Net-based Road Lane Markings Detection for Autonomous Driving

LA Tran, MH Le
ICSSE 2019, IEEE

(see full list at GoogleScholar)





Blog

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 5 min read

POCS-based Clustering Algorithm Explained

Each data has different importance.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 6 min read

EDN-GTM: Encoder-decoder Network with Guided Transmission Map

When traditional computer vision and deep learning conjugate.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 4 min read

MetaFormer: De facto need for Vision?

What is the key to the success of Vision Transformers?

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 5 min read

Data Distillation for Object Detection

Learn from different perspectives.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 6 min read

Learning Without Forgetting Simplified

Learn once and forever.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 5 min read

Data Augmentation Compilation with Python and OpenCV

Prevent your model from overfitting.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 4 min read

Darkeras: Execute YOLOv3/YOLOv4 Object Detection on Keras with Darknet Pre-trained Weights

Everything in the universe is connected.

HTML5 Bootstrap Template by colorlib.com
Towards Data Science | 4 min read

Synthesize Hazy/Foggy Images using Monodepth and Atmospheric Scattering Models

When fog is beneficial.

HTML5 Bootstrap Template by colorlib.com
Towards AI | 5 min read

A Gentle Introduction to Hint Learning & Knowledge Distillation

Guide a student to learn a teacher’s behavior.


*More posts can be found at my Medium page.