Liang-Chieh (Jay) Chen- Home Page
 |
Liang-Chieh Chen Email: lcchen at cs dot ucla dot edu |
About me
Since February 2016, I have been working at Google, Los Angeles.
I finished my Ph.D. in Computer Science at UCLA under the supervision of Alan L. Yuille in November 2015.
My research interests include Computer Vision, Graphical Models, and Machine Learning.
News
05/08/2019: MobileNetV3 is on arxiv now!
04/24/2019: Google blog about running DeepLabv3+ on Google Cloud TPU.
03/07/2019: Evaluation code for image parsing (aka. panoptic segmentation) in terms of Parsing Covering and re-implemented Panoptic Quality is open-sourced!
02/27/2019: FEELVOS (fast end-to-end video object segmentation model) in TensorFlow is open-sourced!
02/26/2019: Auto-DeeoLab model definition in TensorFlow has been open-sourced (called hnasnet in the code) along with PNASNet.
An implementation and checkpoint for the best Dense Prediction Cell trained on Cityscapes have been released, which attains 82.7% on the test set.
DeepLabv3 checkpoints with MobileNet-v2 depth-multiplier=0.5 (pretrained on COCO and PASCAL VOC 2012) have been released, which targets at faster inference speed.
DeepLabv3+ was featured on Computer Vision News.
One paper (neural architecture search for dense image prediction) is accepted to NeurIPS 2018.
Two papers (DeepLabv3+ and PersonLab) are accepted to ECCV 2018.
Check out our latest work, PesonLab, on joint person instance segmentation and pose estimation.
Google AI blog about DeepLabv3+ open source.
Our DeepLabv3+ implementation in TensorFlow is publicly available now!
Education
Ph.D., Computer Science, University of California- Los Angeles, September 2010 - November 2015
M.S., Electrical Engineering: Systems, University of Michigan- Ann Arbor, September 2006 - June 2008
Major: Signal Processing, Minor: Communication
B.S., Electronics Engineering, National Chiao-Tung University, Taiwan, September 2000 - June 2004
Papers
 |
Searching for MobileNetV3 |
 |
FEELVOS: Fast End-to-End Embedding Learning for Video Object
Segmentation |
 |
DeeperLab: Single-Shot Image Parser |
 |
Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation (oral) |
 |
The Devil is in the Decoder: Classification, Regression and GANs |
 |
Searching for Efficient Multi-Scale Architectures for Dense Image
Prediction |
 |
PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model |
 |
Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation |
 |
MobileNetV2: Inverted Residuals and Linear Bottlenecks |
 |
MaskLab: Instance Segmentation by Refining Object Detection with Semantic
and Direction Features |
 |
The Devil is in the Decoder |
 |
Rethinking Atrous Convolution for Semantic Image Segmentation |
 |
DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs |
 |
Zoom Better to See Clearer: Human Part Segmentation with Auto Zoom Net |
 |
Semantic Image Segmentation with Task-Specific Edge Detection Using CNNs and a Discriminatively Trained Domain Transform |
 |
Attention to Scale: Scale-aware Semantic Image Segmentation |
 |
ABC-CNN: An Attention Based Convolutional Neural Network for Visual Question Answering |
 |
Weakly- and Semi-Supervised Learning of a Deep Convolutional Network for
Semantic Image Segmentation |
 |
Learning Deep Structured Models (oral) |
 |
Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs |
 |
Beat the MTurkers: Automatic Image Labeling from Weak 3D Supervision |
 |
Modeling Image Patches with a Generic Dictionary of Mini-Epitomes |
 |
Learning a Dictionary of Shape Epitomes with Applications to Image Labeling |
US Patents
Systems and Methods for Data Page Management of NAND Flash Memory Arrangements, November 2008
Liang-Chieh Chen and Xueshi Yang
Multi-Mode Encoding for Data Compression, February 2009
Liang-Chieh Chen and Xueshi Yang