Welcome to my portfolio.
What do I do for a living?

Computer Vision is my thing. I build and fine-tune custom Deep Learning models from different DL frameworks and deploy it on JVM and pythonic systems. I do research on state-of-the-art deep learning models and perform model migration, optimization and deployment on local servers, mobile devices, cloud servers and embedded systems. I also train engineers from the industry on deep learning fundamentals and practical skills.
我擅长于深度学习模型调参与迁移学习。 模型与Java,Python,C++软件系统整合。 物体检测与分类,精通于计算机视觉。

Recent Work

Person + Skin Segmentation - November 4 , 2020

Person + Skin Segmentation
Person Segmentation + Skin Segmentation + Calculate Skin to Image Size Ratio
This project uses PyTorch, OpenCV
Code is Here

Online Classroom Attention Tracker - July 16 , 2020

Online Classroom Attention Tracker
Monitor student's attention metrics
This project uses PyTorch, OpenCV, PyQtGraph, ZeroMQ, React
Code is Here

Social Distance Monitoring System in Java - April 30 , 2020

Social Distance Monitoring System
Features include social distance monitoring, people counting.
This project uses DL4J, JavaCV, Keras, built using Java.
Code is Here
Youtube video is Here

Blackjack Robot - March 3 , 2020

Autonomous Blackjack Playing Robot
Features include Card Recognition,Poker Chip Recognition, Auto-Dealing, Orientation Correction.
This project uses DL4J, JavaCV, Keras, built using Java.
Youtube video is Here

AI Fitness Counter - November 26 , 2018

Demo of Exercise Counter in 2018 Penang Science Fair, tested by actual students.
GymmyFive Coming Soon in 2020 with more Deep Learning in it!
Application uses Intel OpenVINO toolkit and OpenCV in C++.
Code is Here

MNIST Classifier on Android - September 5 , 2019

Train and Inference a MNIST Classifier on a Android mobile device.
MNIST dataset loaded in Android device and start training on arm CPU using DL4J in Java.
Code is Here
Youtube video is Here

Facial Landmark Prediction - January 5 , 2020

Predicts 68 facial landmarks points by using Dlib library in C++ or Python.

YOLOv2 Object Detection - December 2 , 2019

Inference a pretrained COCO-YOLOv2 model on a video.
Application uses DL4J(Deeplearning4J), JavaCV, FFMpeg in Java.
Code is Here
Youtube video is Here

Get In Touch

Contact me on LinkedIn or Github if you would like to speak to me.