About

I am an incoming student of MSCV (Computer Vision) program at Carnegie Mellon University. Currently I am working with Prof. Kris Kitani. Previously at Inventec Corporation, I was an AI Research Engineer working with Dr. Wei-Chao Chen and Dr. Trista Chen. Before that, I have developed visual search solutions for smart eCommerce at Viscovery. My research interests include deep learning and computer vision, especially anomaly detection, image segmentation and object detection and their applications to smart manufacturing. Also I am fortuante to have worked with Prof. Min Sun. I received my B.S. from National Tsing-Hua University. I am actively looking for 2021 Summer Internship to explore more computer vision applications in the industry.

Carnegie Mellon University

M.S. in Computer Vision

Start in 2021 Spring

Inventec Corporation

AI Research Engineer

2019.01- 2020.09

Viscovery

Computer Vision Engineer

2018.04-2018.12

NTHU

B.S. in PME

2013.09-2017.06

Invited Talk

NVIDIA GTC 2020
Toward Taming the Training Data Complexity in Smart Manufacturing
Trista Pei-Chun Chen, Yi-Chun Chen

Publications

TrustMAE: A Noise-Resilient Defect Classification Framework using Memory-Augmented Auto-Encoders with Trust Regions
Daniel Stanley Tan, Yi-Chun Chen, Trista Chen, Wei-Chao Chen

WACV 2021
Towards 360° Show-and-Tell
Shih-Han Chou, Yi-Chun Chen, Cheng Sun,
Kuo-Hao Zeng, Ching-Ju Cheng, Jianlong Fu, Min Sun

ECCVW 2018
DLWV2: a Deep Learning-based Wearable Vision-system with
Vibrotactile-feedback for Visually Impaired People to Reach Objects
Meng-Li Shih, Yi-Chun Chen, Chia-Yu Tung, Cheng Sun,
Ching-Ju Cheng, Liwei Chan, Srenivas Varadarajan, Min Sun

iROS 2018
Self-view Grounding Given a Narrated 360° Video
Shih-Han Chou, Yi-Chun Chen, Kuo-Hao Zeng,
Hou-Ning Hu, Jianlong Fu, Min Sun

AAAI 2018

Projects

Crack detection on concrete surface
- I implement and analyze various image segmentation models’ capability to detect cracks in concrete images.
- I devise an evaluation metric called kernel-based average precision that overcomes the bias of standard segmentation metrics against thin segments.