About meAbout me
My Timeline
MAR 2021 - FEB 2023
Computer Vision - Ulsan University, Korea
- Masters Degree in Computer Vision
- Graduating Feb 2023
- Currently writing a thesis about real time semantic segmentation with an efficient attention network based on SENetwork and CBAM
SEP 2020 - JUN 2022
Software Engineer - CodePro, Korea
- Developed various machine learning applications utilizing computer vision CNN models for object detection, semantic segmentation and action recognition.
SEP 2019 - MAR 2021
Web Developer - White Cloud Apps
- Developed a RESTful API using Node.js/Express for sending fundrasing events info and users data to email server.
- Created configurable React components to display different event information.
- Developed a responsive design for different screen resolutions using Bootstrap and CSS media queries.
FEB 2019 - MAR 2020
42 Silicon Valley - Coding Bootcamp
- A college-level, non-profit, tuition-free computer programming school with a peer-to-peer learning environment.
- Created unix basic function(printf, ls etc) in C
- Created Tinder like dating, Twitter like social media applications with MERN stack
My ProjectsMy Projects
![](img/resized_port1.png)
Pedestrian Guidance System
- Developed an augmented reality prototype application
- Visualized the user location and information on the screen with a map API
- Trained and applied the STDC2 real time semantic segmentation model
- Preprocessed a road/sidewalk image dataset of 50,000 labeled images
![](img/resized_port2.png)
Realtime Coast Guard Monitoring System
- Developed a real time monitoring system with YOLO v5 object detection model.
- Made system concurrently infer input data and visualize the result.
![](img/resized_port3.jpg)
Illegal Construction Detection
- With the help of Deformabel Convolutional instance segmentation, detects and labels buildings in highly densed city environments.
- Compared temporal result of the instance segmentation, marked different buildings with an accuracy of 34%
![](img/resized_port4.png)
Animal Action Recognition
- Developed an animal action recognition system with TSN CNN action recognition model
- Preprocessed 98,313 animal action videos, each of a duration between 15 - 30 seconds
- Achieved 82% top1 accuracy with 21 action categories of cats and dogs
Contact MeContact
: Columbus, OH
: edgarcjeong@gmail.com
: (614) 615-4082
: Korean, English