Online Learning

For practical applications such as robotics or video analytics,it is necessary to have online self-learning capabilities, and traditional methods based on homogenous "Batch" training cannot meet requirements. We intend to propose an online learning method and design an online measurement learning model. At the same time, we eliminate feature redundancy and rank redundancy of data, avoid overfitting problems, reduce model dimensions, and improve detection efficiency.

Slected publications：

Yang Cong, Baojie Fan, Ji Liu, Jiebo Luo,Haibin Yu, Speeded up Low Rank Online Metric Learning for Object Tracking, IEEE Transactions on Circuits and Systems for Video Technology, 2015.

Yang Cong, Ji Liu, Junsong Yuan, Jiebo Luo, Self-supervised Online Metric Learning with Low Rank Constraint for Scene Categorization,IEEETransactions on Image Processing, v22, n 8, pp 3179 - 3191, 2013.

For public safety, aiming at the multi-person flow density area in public places,we use video image analysis and identification technology, and has developed an automatic identification system for video line and area dynamic flow density, which can automatically identify and calculate dynamic line crossings and regional flow density from multiple angles in real time.

Slected publications：

Yang Cong, Haifeng Gong, Songchun Zhu, Yandong Tang "Flow Mosaicking: Real-time Pedestrian Counting without Scene-specific Learning", IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), p 1093-1100, 2009.

Yang Cong, Ji Liu, Junsong Yuan, Jiebo Luo, Low-Rank Online Metric Learning, Page 203-233, in Yun Fu, Low-Rank and Sparse Modeling for Visual Analysis, Springer, 2014 ISBN: 978-3-319-11999-1 (Print) 978-3-319-12000-3.

Abnormal behavior detection for public safety

Medical image analysis and navigation systems can effectively reduce the workload of doctors and increase the recognition rate of lesions. A depth feature selection model based on sparse constraints is proposed to reduce the impact of redundant and disturbance information in high-dimensional features and improve the processing speed.

Our DataSet：Medical Image Analysis Dataset


Slected publications：

Yang Cong, Shuai Wang, Ji Liu, Jun Cao, Yunsheng Yang, Jiebo Luo, Deep Sparse Feature Selection for Computer Aided Endoscopy Diagnosis, Pattern Recognition, Volume 48, Issue 3, March 2015, Pages 907-917.

Shuai Wang, Yang Cong, Huijie Fan, et al., "Multi-class Latent Concept Pooling for Computer-Aided Endoscopy Diagnosis", ACM Transactions on Multimedia Computing, Communications, and Applications, Volume 13 Issue 2, 2017.

Medical Image Analysis and Navigation

Industrial big data processing

For Industry 4.0 and Smart Manufacturing 2025, the industry has entered the "big data" era. Our research focuses on solving the problem of "data missing" industrial big data processing caused by acquisition, transmission, cost, and condition constraints, and proposes a multi-task machine learning model based on low-rank constraints, which can utilize limited acquisition information and achieve Multi-task prediction, classification, and other industrial data processing needs.

Slected publications：

Yang Cong, Baojie Fan, Ji Liu, Jiebo Luo,Haibin Yu, Speeded up Low Rank Online Metric Learning for Object Tracking, IEEE Transactions on Circuits and Systems for Video Technology, 2015.

Yang Cong, Ji Liu, Baojie Fan, Haibin Yu, Jiebo Luo, Online Similarity Learning for Big Data with Overfitting, IEEE Transactions on Big Data, 2017.

New General Machine Vision

With the shortage of labor and the increase in costs, the manufacturing industry is in urgent need of upgrades in production, with machines replacing people. One of the key technologies that needs urgent breakthrough is a more intelligent and flexible machine vision system. We have designed a machine vision system with on-line self-learning capability that can adapt to 2D and 3D environments. We have achieved any workpiece recognition, matching, and pose estimation. This system can be used for rapid sorting, palletizing, intelligent assembly, and positioning of industrial robots. , defect detection and other fields.

Slected publications：

Yang Cong, Haifeng Gong, Yandong Tang, Shuzhi Sam Ge, Jiebo Luo, Real-time One-dimensional Motion Estimation and its Application in Computer Vision, Machine Vision and Application, v26, issue 5, p633-648, 2015.

Yang Cong, Junsong Yuan, Yandong Tang, Video Anomaly Search in Crowded Scenes via Spatio-Temporal Motion Context, IEEE Transactions on Information Forensics & Security, v 8, issue 10, pp 1590-1599, October 2013.

Shenyang Institute of Automation Chinese Academy of Sciences 1996-2013   LiaoICP 05000867