Jinjin GU

CUHK(SZ), The Lab of Energy Internet, Research Assistant.
Sensetime Research, Research Intern.

CUHK(SZ), ChengDao Bldg, 204

hellojasongt AT gmail.com, gujinjin AT sensetime.com
CV / GitHub / Google Scholar / LinkedIn

The Lab of Energy Internet, CUHK(SZ), Sensetime Research


I am currently studying in the School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen (CUHKSZ) and my major is Computer Science and Engineering (CSE). I am a research assistant of the Lab of Energy Internet, under the supervision of Prof. Junhua Zhao. I am also a research intern at the Sensetime Research hosted by Prof. Liang Lin and Dr. Chao Dong. Before that, I was a research assistant at the Institute of Image Communication and Network Engineering in Shanghai Jiao Tong University (June to August, 2017), supervised by Prof. Xiaolin Wu.


Research Statement

My research interests lie primarily in the theory and application of machine learning, including the representation learning, manifold learning and the application of information geometry in machine learning. I am also interested in the application of machine learning approaches in computer vision. Including learning-based image and video processing, image and 3D segmentation and visual reasoning. I am trying to be involved in more research areas as much as possible, especially in the interdisciplinary field of advanced information technology and other areas, hoping to find the academic goals that can struggle for life.

Research Overview

Computer Vision and Image Processing
  • Image Super-resolution, Denoising and Deblur.
  • Applications of Generative Models.
Machine Learning in Industrial Applications
  • Super Resolution Perception of Industrial Sensor
  • Data Driven Electricity Economics
Deep Learning Theory
  • Interpretability of Deep Learning Models.


ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
Talk at ECCV 2018 Workshop and Challenge on Perceptual Image Restoration and Manipulation
We thoroughly study three key components of SRGAN - network architecture, adversarial loss and perceptual loss, and improve each of them to derive an Enhanced SRGAN (ESRGAN).


Computer Vision and Computer Graphics

A Simple Yet Effective Baseline for Silky Hair Images Generation from Sparse Sketches
Haonan Qiu, Xiaoguang Han, Hang Zhu, Jinjin Gu, Peng Liu
AAAI 2019 Under Review
We explore GANs in hair generation and put forward HairGAN. Our model consists of two convolutional networks and one domain mapping module, trained in a supervised fashion on pairs of real-like hair images and corresponding sparse sketches.
ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Chen Change Loy, Yu Qiao, Xiaoou Tang
ECCV 2018 Workshop
We thoroughly study three key components of SRGAN - network architecture, adversarial loss and perceptual loss, and improve each of them to derive an Enhanced SRGAN (ESRGAN).
Single Image Reflection Removal Using Deep Encoder-Decoder Network
Zhixiang Chi, Xiaolin Wu, Xiao Shu, Jinjin Gu
arXiv preprint
We propose a novel deep convolutional encoder-decoder method to remove the objectionable reflection by learning a map between image pairs with and without reflection.

Machine Learning in Industrial Applications

Spotlight Super-Resolution Perception of Industrial Sensor
Jinjin Gu, Guolong Liu, Gaoqi Liang, Junhua Zhao
AAAI 2019 Under Review
we present the problem formulation and methodology framework of Super-Resolution Perception (SRP) on industrial sensor data. This technology makes it possible for empowering existing industrial facilities without upgrading existing sensors or deploying additional sensors.


Spotlight SenseSR - Intelligent Photography Solution for Mobile Devices
Research at Sensetime Research
Denoising and super-resolution of multiple unaligned observation pictures with unknown noise and unknown blur under the limited computing conditions and time constraints on mobile devices.


Flag Counter