Yvon Wong
PostDoc at Academia Sinica
Taiwan
Summary
🤩
Rockstar🎓
Top SchoolYvon Wong is a PostDoc and machine learning engineer based in Taiwan with nine years of experience specializing in real-time object detection and model engineering. He has made significant open-source contributions to prominent YOLO implementations (including YOLOv7 and YOLOv4), focusing on core architecture, loss functions, target building, reparameterization, and dataset pipelines for COCO. His work blends research rigor with practical tooling—improving demo/visualization scripts and data preparation to accelerate training and reproducibility. At Academia Sinica he continues to bridge academic research and production-ready code, demonstrating an uncommon combination of deep model internals and hands-on engineering. Colleagues describe him as someone who improves both model performance and developer experience, often by addressing non-obvious plumbing like anchor configuration and training utilities.
9 years of coding experience
Chien Kuo High SchoolNSYSU
Github Skills (23)
pytorch10
loss-functions10
python10
data-set10
machine-learning10
datasets10
data-preprocessing10
darknet10
dataprep10
trainings10
object-detection10
computer-vision10
yolov710
modeling10
yolov410
Programming languages (4)
Github contributions (5)
WongKinYiu/yolor
Apr 2021 - Mar 2022
implementation of paper - You Only Learn One Representation: Unified Network for Multiple Tasks (https://arxiv.org/abs/2105.04206)
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 149 commits, 114 pushes in 10 months
Contributions summary:Yvon contributed to the project by creating scripts to download and prepare datasets for training, specifically for COCO. They added and modified files, including Python scripts, related to data loading, dataset processing, and model definition. Furthermore, they created files for training and testing the models. Their work included modifying a `metrics.py` file.
pytorchrepresentationarxivimplicitabs
WongKinYiu/yolov7
Jul 2022 - Dec 2022
Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
Role in this project:ML Engineer
ML EngineerContributions:1 release, 120 commits, 132 pushes in 5 months
Contributions summary:Yvon primarily contributed to the core code of the YOLOv7 project, implementing and modifying key components for object detection and keypoint estimation. Their work focused on loss functions, target building, and model reparameterization, indicating a deep involvement in the model's architecture and training process. They also updated demo scripts and visualization tools, enhancing the usability and demonstrability of the model.
pytorchreal-timeartdeep-learningobject-detection
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.