Xiaopeng Yan

    I am a master student in Human Cyber Physical Intelligence Integration Lab at the SUN YAT-SEN UNIVERSITY, led by Professor Liang Lin. Now, I am majoring in computer science at School of Data and Computer Science during M.S. I received a B.E. in automation at School of Electronics and Information Technology. Now, I'm as an intern at SenseTime from July,2018.


Download CV Here


Computer Vision
object detection
video classification
generative adversarial network (GAN)
Machine Learning
self-supervised learning and deep active learning
weakly-supervised learning


Towards Human-Machine Cooperation:Self-supervised Sample Mining for Object Detection
Keze Wang,Xiaopeng Yan, Dongyu Zhang, Lei Zhang, Liang Lin
CVPR, 2018  
Project / PDF / code / pytorch-version

Though quite challenging, leveraging large-scale unlabeled or partially labeled images in a cost-effective way has increasingly attracted interests for its great importance to computer vision. To tackle this problem, many Active Learning (AL) methods have been developed. However,these methods mainly define their sample selection criteria within a single image context, ...


Cost-Effective Object Detection: Active Sample Mining with Switchable Selection Criteria
Keze Wang, Liang Lin, Xiaopeng Yan, Dongyu Zhang, Ziliang Chen, Lei Zhang
IEEE Transactions on Neural Networks and Learning System (TNNLS), 2018  
Project / PDF / code / pytorch-version

Though quite challenging, the training of object detectors using large-scale unlabeled or partially labeled datasets has attracted increasing interests from researchers due to its fundamental importance for applications of neural networks and learning systems. To address this problem, many active learning (AL) methods have been proposed that employ upto-date detectors ...


Self-Learning Framework for Visual Recognition
Excellent graduation thesis of SYSU, 2017  
PDF / code

In the graduation paper a general self-learning framework is proposed. It is effective on semi-supervised object detection. The proposed framework proposes to train object detectors by faithfully recognizing high-confidence object proposals in a self-paced way, and discovering low-confidence ones for user annotation in a active learning way.


Running Hi
Guoen Tao,Xiaopeng Yan,Hao Wang,Xin Zhang,Ziyun Zhang,Yongqiu,Yishun Zheng,Jingsen Chen
Sun Yat-sen University & Guangzhou University, 2014-2015  
PDF / Award / code

This project aims to develop an App and a Websit to supply service for college students to run and social interaction.We got the first place in the south china division of Lenovo Group national students entrepreneurship competition.


As an intern in China Telecom
Guang Zhou, 2015.7-8  

During the internship, I almost used the big data paltform such as Hadoop and Spark to analysis the users' data. By analysising the users' all kinds of data, We maked a plan where to bulid a billboard on the freeway.


As a developer in the Lab for big data and communication in SYSU
Guang Zhou, 2014.8-2015.6  
code of Weibo, code of LDA

During these time, I writed a crawler program to crawl the Weibo users' data by the user's location. And I also writed a Latent Dirichlet Allocation algorithm to analysis the users' data in the form of HTML.