Bio
Kunbo Zhang is an Associate Professor at New Laboratory of Pattern Recognition (NLPR), State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences (CASIA). He is also the Deputy Director of Tianjin Academy for Intelligent Recognition Technologies (TAFIRT). He was a Post-doc in NLPR CASIA. Prior to joining CASIA, he worked as a Manufacturing Engineer in Nexteer Automotive (Michigan, U.S.). He received his B.A. in Automation from Beijing Institute of Technology, and his M.S. and Ph.D. from State University of New York at Stony Brook.
张堃博，副研究员，中国科学院自动化研究所、多模态人工智能系统全国实验室、模式识别北京市重点实验室，本科毕业于北京理工大学自动化专业，硕士和博士毕业于纽约州立大学石溪分校机械工程专业，博士后于模式识别国家重点实验室工作，曾于2011-2016年于Nexteer Automotive（原通用汽车）任职，近五年主持国家自然科学基金面上项目、国家科技重点专项、国家重点研发计划课题与子课题、科技部通用人工智能技术平台以及技术负责包括国家重大科研仪器研制项目等二十余项国家和省部级项目，主持四十余项企业委托项目，累计发表期刊和会议文章 40 篇，授权和申请国家发明专利和PCT专利 33 项，获得软著 11 项，参与制订国家标准 6 项，所负责智能系统项目部署于世界四大洲，项目总额超 3 亿元。
Research Interests
His research interests focus on Computer Vision, AI4Science, AI Applications
主要研究方向为计算机视觉、AI4Science、人工智能系统应用
Research Projects
Research Directions



Current Projects

  1. Tumor Diagnosis using Multi-channel Femtosecond Image | 基于多通道飞秒图像的肿瘤良恶性诊断分析
    Yunlong Wang, Mengmeng Cui, Zhuofeng Zhou, Kunbo Zhang
    Our self-evolution method for breast cancer diagnosis has reached the AUC=0.9934 using only 130 samples. Its generalization and new bio-marker discovery capabilities are also demonstrated by multi-center data from several provinces. We are working on a new pipline for better unerstanding of the information introduced by Femtosecond and exploring the possiblity of better diagnosis of multiple diseases over traditional image knowledge paradigm.

  2. Detection of Academic Misconduct in Research Papers | 科研论文学术不端检测技术与监测平台
    Kunbo Zhang, Mengmeng Cui, Yuanzu Hu, Leyuan Wang, Yiwei Ru, Zhenteng Shen, etc.
    We have developed a series of AI methods and built up a system to detect image forgery, image and text generations, and semantic plagiarism, which run on an extremly large vector database of millions of academic papers. This system is under trial operation.

  3. Robust Face Anti-spoofing from Cross-modal Visual Clues | 基于跨模态视觉线索的鲁棒人脸活体检测方法
    Tian Yu, Yalin Huang, Kunbo Zhang
    We tackle the face anti-spoofing problem from the angle of physics by using of multi-modal visual information. To overcome the drawbacks of prevous methods, the knowledge of light intensity and light osillation angle are introduced for a thorough understanding of the difference between genuine face and fake attacks.

  4. High-throuput Iris Recognition System | 高通量虹膜识别系统
    Kunbo Zhang, Zhenteng Shen, Yunlong Wang
    To develop next generation iris recognition system, we are working on multi-modal biometric imaging devices, recognition algorithm, and integrated system together. High-throughput system requires extremly efficient and accurate biometric information capture and handling. Our new sytem is able to recognize multiple -person iris in an extended capture volume (up to 3 persons at 10 m distance) and accomplish single-person iris-on-the move recognition. Currently we are developing and testing version 3.0 system.

  5. Extended DoF Eye-tracking Equipment and Applications in Human-machine Interaction | 景深扩展的眼动追踪设备及人机交互应用
    Kunbo Zhang, Zhenteng Shen, Yunlong Wang
    We design and develope the innovative binocular long distance eye tracking equipment.The focus-tunbale lens is used to extended the depth-of-field of traditional eye tracking device. The fast response and precise control imaging setup allows to track two-pair of human eyes over 1.5 m distance simultaneously.
News
Grants
Current Grants
Past Grants
Selected Papers