Avatar

Jiarui Xu

Undergraduate Student of HKUST / Research Intern at MSRA

Biography

Jiarui Xu is currently an undergraduate student in Hong Kong University of Science and Technology. He was a visiting student in University of California San Diego mentored by Hao Su during 2019 Summer. He was also a full-time research intern in Visual Computing Group of Microsoft Research Asia from May, 2018 to April 2019, supervised by Yue Cao, Han Hu, Stephen Lin. Prior to that, he worked with Yu-Wing Tai and Chi-Keung Tang as a research undergraduate.

His research interests currently focus on computer vision and machine learning. He has experiences in image classification, object detection, action recognition and point cloud segmentation.

He is expected to complete B.S degree in Computer Science and Mathematics by 2020 Spring and applying for Ph.D starting from 2020 Fall. Here is his Curriculum Vitae.

Interests

  • Computer Vision
  • Robotics
  • Machine Learning
  • Artificial Intelligence

Education

  • Visiting Undergradute in SU Lab, 2019 Summer

    University of California San Diego

  • Research Intern in Visual Computing Group, 2018 - 2019

    Microsoft Research Asia

  • Exchange Student in School of Engineering, 2018 Spring

    Washington University in St.Louis

  • BSc in Computer Science & Mathematics, 2015 - 2020

    Hong Kong University of Science and Technology

News:

  • 2019.10 Our paper GCNet won the Best Paper Award at ICCV 2019 Neural Architects Workshop!
  • 2019.09 One paper got accepted by ICCV Neural Architects Workshop 2019 as Oral
  • 2019.07 One paper got accepted by ICCV 2019
  • 2019.06 GCNet is merge into official repo of mmdetection github
  • 2019.04 Code of GCNet github released.
  • 2019.01 DCN ops github with Pytorch 1.0.0 support repleased.
  • 2018.07 One paper got accepted by ECCV 2018

Recent Publications

* indicates equal contribution

GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond.
Internation Conference on Computer Vision Workshop on Neural Architects (ICCVW) 2019 Best Paper Award, submitted to TPAMI Special Issue.
Yue Cao*, Jiarui Xu*, Stephen Lin, Fangyun Wei, Han Hu

PDF Code Code@mmdet Post

Spatial-Temporal Relation Networks for Multi-Object Tracking.
In Internation Conference on Computer Vision (ICCV) 2019.
Jiarui Xu, Yue Cao, Zheng Zhang, Han Hu

PDF

End-to-End Deep HDR Imaging with Large Foreground Motions.
In European Conference on Computer Vision (ECCV), 2018.
Shangzhe Wu, Jiarui Xu, Yu-Wing Tai, Chi-Keung Tang

PDF Code Poster Supplementary

Featured

March 2019 Internation Conference on Computer Vision Workshop on Neural Architects (ICCVW) 2019 Best Paper Award, submitted to TPAMI Special Issue

GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond

  • Re-interpretation and connection of two top popular techniques proposed in CVPR 2018 (Non-local neural networks and SE-Net).
  • Unification of these the two techniques into a general framework.
  • Better instantiation of the general framework, which is about 50x faster than the non-local neural block, while achieving better accuracy than both techniques (non-local and SE-Net) on several recognition tasks such as ImageNet classification, COCO object detection and Kinetics action recognition.

PDF Code Code@mmdet Post
March 2019 In Internation Conference on Computer Vision (ICCV) 2019

Spatial-Temporal Relation Networks for Multi-Object Tracking

  • The first coherent and end-to-end framework for similarity measure which combines all of the appearance, motion and interaction cues.
  • Properly redesign of feature representation for the tracklet-object pair.
  • Achieve the state-of-the-art multi-object tracking (MOT) results on all of the MOT15-17 leaderboards using few bells and whistles.

PDF
March 2018 In European Conference on Computer Vision (ECCV), 2018

End-to-End Deep HDR Imaging with Large Foreground Motions

This paper proposes the first end-to-end deep framework for high dynamic range (HDR) imaging of dynamic scenes with large-scale foreground motions.

PDF Code Poster Supplementary

Projects

Video Object Segmentation

Fast Video Object Segmentation with Temporal Aggregation Network and Dynamic Template Matching

RoboMasters

Champion in International Wildcard Game
detail video

Working Experience

 
 
 
 
 

Research Intern

Microsoft Research Asia

May 2018 – Apr 2019 Beijing, China
  • Researched on image classification, object detection and tracking, action recognition.
  • Implemented and maintained group code bases and infrastructures.
  • Served as an external contributor of mmdetection with kernel operators like DCN, GCNet.
  • Completed research projects submitted to conferences and journals.
 
 
 
 
 

Software Developer Intern

Magnum Research Limited

Sep 2017 – Jan 2018 Hong Kong, China
  • Built both front end and back end infrastructures for quant trading.
  • Developed, beatify and maintain the company main web page.
  • Designed and built the company resource CMS system.
 
 
 
 
 

Software Developer Intern

TFI Digit Limited

Jun 2017 – Aug 2018 Hong Kong, China
  • Implemented the automatic subtitle system for product team.
  • Integrated various speech to text recognition APIs into video annotation pipeline.
  • Developed Meeting tagging iOS application for automatic meeting minutes.
 
 
 
 
 

Senior Embed/Vision SystemDeveloper

HKUST Robotics Institute

Nov 2016 – Jan 2018 Hong Kong, China
  • Won the champion in international wildcard game.
  • Implemented control algorithm of 5 infantry robots for operators.
  • Developed target recognition system for infantry robot with OpenCV.

Visual Computing Group in Microsoft Research Asia (2018-2019)

SU Lab in University of California San Diego (2019 Summer)

RoboMasters Team of Robotics Insitution in Hong Kong University of Science and Technology (2016 - 2017)

CK research group in Hong Kong University of Science and Technology (2017)

Exchange in Washington University in St.Louis (2018 Spring)

Class of 2019 Graduation Dinner

Previous Next

Contact