Close

Yang Jiao

AI Research Engineer, Ph.D.




About

Yang Jiao is an research engineer of artifical intelligence (AI), with research interest in the filed of computer vision, deep learning, neural networks, particulary in digital image processing (low-level) and image/video understanding (high-level) and their applications in optical/scene flow estimation, fine-grained visual categorization (FGVC), multi-modality facial expression recognition (FER), high dynamic range (HDR) and synthetic aperture radar (SAR) image enhancement.

Jiao studied in the School of Electrical Engineering (EE) in Xidian University when he is a undergraduate student. Then he did research in the School of Artifical Intelligence (SAI) and the Department of Electrical and Computer Engineering (ECE) for Ph.D., in Xidian University and the Johns Hopkins University.

He earned his B.S. and Ph.D. degrees in Xidian University in 2015 and 2021 respectively - all in circuit and system, electronical science and technology.

Education

Xidian University

B.S. in Electrical Engineering (EE)

Studied in the school of EE, major with Circuit and System.
Finished course work about science and engineering, did research of video stabilization.

Xidian University

M.S. in Artificail Intelligence (AI)

Studied in the school of EE, major with Circuit and System. Did research about HDR SAR image enhancement and its satellite applications. Adviced by Prof. Guangming Shi and Prof. Yi Niu.

Xidian University

Ph.D. in Artificail Intelligence (AI)

Concentrated on pattern recognition, image understanding, deep learning and neural networks with its applications, e.g. fine-grained visual categorization and multi-modality facial expression recognition. Adviced by Prof. Guangming Shi and Prof. Yi Niu.

Johns Hopkins University

VIS Ph.D. in Electrical and Computer Engineering (ECE)

Concentrated on video understanding, motion consistency anaysis with its deep learning application of optical flow and scene flow estimation. Adviced by Prof. Trac D. Tran.

Experience

Xidian Undergraduate Student Union

2011 - 2014

Joined Xidian University Student Union in 2011 and hosted campus activities as the vice president & head of Dept. of Multi-Media Technology.

Optoelectronic Imaging and Brain-Inspired Perception Laborotary
(OIBP)

2015 - 2021

Joined OIBP in 2015 as a fresh graduate student and studied for 6 years. Love all the members here and appreciate them for the contribution.

Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education (IPIU)

2016 - 2021

 
Research on neural networks, image understanding.

Scholarship Under the State Scholarship Fund (by CSC)

2019 - 2021

Awarded a scholarship under the State Scholarship Fund to pursue Ph.D. study in the U.S.. The awardee was selected through a rigid academia evaluation process organized by China Scholarship Concil (CSC).

Xidian Guangzhou Institute of Technology

2021

Research on optical flow, robotics.

Huawei Technologies Co., Ltd.

2022 - present

Research on deep learning, multi-media for HMS.

Papers & Projects

EffiScene: Efficient Per-Pixel Rigidity Inference for Unsupervised Joint Learning of Optical Flow, Depth, Camera Pose and Motion Segmentation

We address the challenging unsupervised scene flow estimation problem by jointly learning four low-level vision sub-tasks: optical flow F, stereo-depth D, camera pose P and motion segmentation.

Cite           View Paper

EmotionUI

A software for multimodalty 2D+3D facial expression recognition (FER). EmotionUI provides a user friendly interface for solving real-time FER task, including a full technique pipeline, e.g. face detection, pre-processing, 2D FER, 3D FER inference. And it supports customize data collection & training.

View Project

Attention Shift based Deep Neural Network for Fine Grained Visual Categorization

We propose a novel end-to-end FGVC network structure named Attention-Shift based Deep Neural Network (AS-DNN) to locate the discriminative re- gions automatically and encode the semantic correlations iteratively for FGVC task.

Cite           View Paper

Dynamic Range Reduction of SAR Image via Global Optimum Entropy Maximization With Reflectivity-Distortion Constraint

We introduce a new SAR image visualization algorithm to map the high dynamic range SAR amplitude values to low dynamic range displays via reflectivity distortion preserved entropy maximization.

Cite           View Paper

Optical Flow Estimation via Motion Feature Recovery

We discover the Vanishing Cost Volume Problem in optical flow, and propose a novel iterative Motion Feature Recovery (MFR) method to address the the problem via modeling motion consistency across multiple frames.

Cite           View Paper

Attention based Convolutional Neural Network for 2D+ 3D Facial Expression Recognition

We propose an advanced facial attention based convolutional neural network (FA-CNN) for 2D+3D FER to address the existing discriminative regions localization problem.

Cite           View Paper

2D+ 3D Facial Expression Recognition via Discriminative Dynamic Range Enhancement and Multi-Scale Learning

We propose a novel Map Generation technique from the viewpoint of information theory, to boost the slight 3D expression differences from strong personality variations, and design Facial Attention for multi-scale learning.

Cite           View Paper

A novel lossless compression framework for facial depth images in expression recognition

We propose a novel efficient lossless compression framework for facial depth images in expression recognition to reduce the storage size and save bandwidth.

Cite           View Paper

The L_infinity constrained global optimal histogram equalization technique for real time imaging

We remodel the histogram equalization tone mapping task based on graphic theory which achieves the global optimal solutions for HDR to LDR imaging.

Cite           View Paper

GF-** Satellite Image Enhancement System

A software for GF-** Satellite Image Enhancement System, including functions such as multi-band super-large image (15k*15k x 5 band) denosing, single/multi-frame super resolution, have removal.

View Project

Patents

  • 人脸表情分类器的训练、人脸表情的识别方法和装置: CN112906629A[P]. 2021.
  • 一种场景流估计、场景流估计模型的训练方法和装置: CN113160278A[P]. 2021.
  • 基于注意力转移机制的细粒度图像分类方法: CN110598029A[P]. 2019.
  • 基于概率统计与图像梯度信息的全局矢量获取方法: CN105263026A[P]. 2018.
  • 基于子空间正交向量的峰电位检测方法: CN105962932A[P]. 2018.
  • 基于无穷范数约束与最大熵原则的色调映射方法: CN104835121A[P]. 2017.

    • Hashtags / Skills

    Get in Touch