Deep Learning in Computer Vision


Computer vision tries to acquire, process, analyze and understand visual data captured by all kinds of sensors from the real world. It is a crossing discipline between computer science and artificial intelligence, attracting a large population of researchers all over the world. Recent advances in deep learning have provided significant improvements in academia. As a result, deep learning has been adopted in all kinds of computer vision tasks and many breakthroughs have achieved in sub-areas, such as image classification, object detection and so on.

This workshop mainly focuses on deep learning in computer vision. We welcome original and high-quality contributions from academia as well as industry on topics including, but not limited to:


l  Deep neural network design for specific vision applications

l  Deep learning for feature representation

l  Deep learning for video surveillance

l  Deep learning for object recognition/retrieval

l  Deep learning for scene understanding

l  Deep learning for face analysis

l  Deep learning for activity recognition

l  Deep learning for biometrics

l  Deep learning for remote sensing

l  Sparse coding in deep learning

l  Transfer learning for deep learning

l  Deep learning theory

l  Deep learning for dimension reduction

l  Performance evaluation of deep learning

l  Optimization for deep learning



Workshop Chair:

Zhong Zhang, Tianjin Normal University,

Shuang Liu, Tianjin Normal University,

Ronghua Zhang, Shihezi University,


Workshop Co-chair:

Baihua Xiao, Institute of Automation, Chinese Academy of Sciences,

Liang Zheng, Australian National University,

Yuan Xie, East China Normal University,

Yinglu Liu, JD AI,

Lihu Xiao, Beijing IrisKing Co., Ltd.,

Rui Jiang, Shanghai Maritime University,



Smart Air Traffic Management


Artificial Intelligence (AI) has been applied to solve a variety of problems faced by air traffic management in many countries. These include the need to increase the predictability of traffic at different phases of flight, improve passenger flows at airports, and enable greater automation of the system.

Smart air traffic management (ATM) promotes operation efficiency, service quality, and management ability by ultilizing AI technologies, e.g., natural language processing, neutral network, and knowledge graph. Specifically, AI leading to automation would enable the definition and application of smart strategies for managing air traffic and ensuring a high degree of air-ground integration. Automation would also allow for certain tasks to be offloaded, enabling pilots and air traffic controllers to focus on safety critical tasks. Moreover, smart ATM uses the new generation of information technology to facilitae extensive information sharing, application, and coordination.

Solicited topics include, but are not limited to:


l  Instruction Recognition and Processing

l  ATM Knowledge Representation and Acquisition

l  System Wide Information Management

l  Intelligent Decision Support System

l  Big Data in ATM

l  Smart Airport

l  Airport Collaborative Decision Making

l  Application of AI in ATM

l  Decision Making based on Weather Information


Workshop Chair:

Mao Yi, State Key Laboratory of Air Traffic Management System and Technology,