Skip to content

ILoveFree2/FVAS

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fudan Video Analysis System (FVAS)

School of Computer Science, Fudan University

FudanVA is a C++ open-source software for detecting,recognizing and tracking certain people in the video.
You give it a face/person dataset and it can find the people in the video and track him/her.

FudanVA is designed to work on monitors' video.

Source code is provided under a BSD style license. OpenCV with ffmpeg/gstreamer plugins and c++11 are required.

Code, demo, and datasets are available at our website
http://omap.fudan.edu.cn/FVAS

Quick Start

copy data/ into this directory
$make
$./fudanvideo_demo
Or open our project in QT

Main Method

Tracking:
Optical flow from Opencv
Face Detection:
HOG+SVM from Opencv
Body/Person Detection:
HOG+SVM from Opencv
Face Recognition:
ASM+LBP from Stasm and Opencv
Person Recognition/Re-Identification:
Mid-level filters coding by our group

Output

Green windows represent body detection
Red windows represent face detection, and number shows the label by face recognition
White windows represent tracking, and number shows the label by person recognition

Reference

  • S. Milborrow and F. Nicolls

  • Active Shape Models with SIFT Descriptors and MARS

  • VISAPP,2014

  • www.milbo.users.sonic.net/stasm/

  • P. Felzenszwalb, R. Girshick, D. McAllester, D. Ramanan.

  • Object Detection with Discriminatively Trained Part Based Models.

  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 32, No. 9, September 2010

  • Discriminatively Trained Deformable Part Models, Release 4

  • http://people.cs.uchicago.edu/~pff/latent-release4/

  • Rui Zhao Wanli Quyang Xiaogang Wang

  • Learning Mid-level Filters for Person-identification

  • Zhang W, Wang X, Zhao D, et al.

  • Graph degree linkage: Agglomerative clustering on a directed graph

  • LIBSVM:

  • Chih-Chung Chang and Chih-Jen Lin, LIBSVM :

  • a library for support vector machines.

  • ACM Transactions on Intelligent Systems and Technology, 2:27:1--27:27, 2011.

  • Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm

About Version 0.10 alpha

Our project has just begun.There are countless works wait us to do.
I am sorry to remain these bugs and consider less about human design.
We hope open-source can bring us advices and companions. We have a long-time plan around video analysis.
Finally, thanks for using.

About Version 0.10 beta

We plan to release beta version this year.
New functions:

  • train model for face and person recognition
  • a simple UI
  • new demo video
  • may be windows version
  • friendly API for opencv

TO DO

  • Body detection based on DPM

  • Real-time Face Detection based on CNN

  • Speed up linear SVM by optimizing matrix computation

  • Tracking based on TLD

Note

  • Contributors: Qipeng Guo, Jiajun Ou, Yanye Li, Zhedong Zheng, Guangzhen Zhou, Fengdong Zhang, Dequan Wang
  • Thanks for helping from BarclayII, David Gao (Github account)
  • School of Computer Science , Fudan university
  • Time: 2014 You can get more information from our website http://omap.fudan.edu.cn/FVAS

Schedule

http://10.141.208.19/schedule.html

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 97.2%
  • C 1.8%
  • Other 1.0%