Introduction

This is a fork of the C++ implementation of OpenTLD called CFtld. OpenTLD was originally proposed in [1] and implemented by Georg Nebehay in C++.

This fork uses C++ implementations of correlation filter based trackers as short-term trackers. Both short-term trackers are modified variants of the visual trackers proposed in [2,3]. The short-term trackers are extended with target loss detection capabilities as in [4] and use the C++ implementation [5] of the FHOG features proposed in [6]. More implementation details of the short-term trackers can be found here. The detection cascade is only used to suggest possible target locations to the short-term trackers for redetection purposes. It cannot reinitialize the short-term trackers. The short-term trackers decide whether a suggested patch actually contains the target.

The system starts with the KCFcpp as default short-term tracker.

Usage

If you have a webcam attached to your PC, you can simply execute cftld (on Linux) or cftld.exe (on Windows) in order to try it out.

Keyboard shortcuts

• r select a target
• q quit
• l toggle learning
• a toggle alternating mode (if true, detector is switched off when tracker is available)

Command line options

Synopsis

cftld [option arguments] [arguments]

option arguments

• -x Use DSSTcpp instead of KCFcpp as short-term tracker. KCFcpp is the default short-term tracker.
• [-a <startFrameNumber>] video starts at the frameNumber startFrameNumber
• [-b <x,y,w,h>] Initial bounding box
• [-d <device>] select input device: device=(IMGS|CAM|VID|STREAM)
  IMGS: capture from images
  CAM: capture from connected camera
  VID: capture from a video
  STREAM: capture from RTSP stream
• [-f] shows foreground
• [-i <path>] path to the images or to the video.
• [-j <number>] show trajectory for the last number frames
• [-h] shows help
• [-n <number>] Specifies the video device to use (defaults to 0). Useful to select a different camera when multiple cameras are connected.
• [-p path] prints results into the file path
• [-s] if set, user can select initial bounding box
• [-t <theta>] threshold for determining positive results
• [-z <lastFrameNumber>] video ends at the frameNumber lastFrameNumber. If lastFrameNumber is 0 or the option argument isn't specified means all frames are taken.

Arguments

[CONFIG_FILE] path to config file

Config files

Refer to the sample folder to see usage and config file examples.

Build

Dependencies

• C++11
• OpenCV 3.0
• CMake
• libconfig++ (provided)
• SSE2-capable CPU

Windows 7

• Set environment variables according to http://docs.opencv.org/doc/tutorials/introduction/windows_install/windows_install.html
• Launch cmake-gui, create a build folder and configure.
• Open CF_TLD.sln in Visual Studio and compile the project opentld.

Ubuntu 14.04

• Install OpenCV 3.0 and CMake.
• Configure and compile:

mkdir <src-dir>/build
cd <src-dir>/build
cmake ../
make -j 8

Commercial Use (US)

The code using linear correlation filters may be affected by a US patent. If you want to use this code commercially in the US please refer to http://www.cs.colostate.edu/~vision/ocof_toolset_2012/index.php for possible patent claims.

References

If you reuse this code for a scientific publication, please cite the related publications (dependent on what parts of the code you reuse):

[1]

@article{kalal2012TLD,
title={Tracking-Learning-Detection},
author={Kalal, Zdenek and Mikolajczyk, Krystian and Matas, Jiri},
journal={Pattern Analysis and Machine Intelligence, IEEE Transactions on},
year={2012}}

[2]

@article{henriques2015tracking,
title = {High-Speed Tracking with Kernelized Correlation Filters},
author = {Henriques, J. F. and Caseiro, R. and Martins, P. and Batista, J.},
journal = {Pattern Analysis and Machine Intelligence, IEEE Transactions on},
year = {2015}

[3]

@inproceedings{danelljan2014dsst,
title={Accurate Scale Estimation for Robust Visual Tracking},
author={Danelljan, Martin and H{\"a}ger, Gustav and Khan, Fahad Shahbaz and Felsberg, Michael},
booktitle={Proceedings of the British Machine Vision Conference BMVC},
year={2014}}

[4]

@inproceedings{bolme2010mosse,
author={Bolme, David S. and Beveridge, J. Ross and Draper, Bruce A. and Yui Man Lui},
title={Visual Object Tracking using Adaptive Correlation Filters},
booktitle={Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2010}}

[5]

@misc{PMT,
author = {Piotr Doll\'ar},
title = {{P}iotr's {C}omputer {V}ision {M}atlab {T}oolbox ({PMT})},
howpublished = {\url{http://vision.ucsd.edu/~pdollar/toolbox/doc/index.html}}}

[6]

@article{lsvm-pami,
title = "Object Detection with Discriminatively Trained Part Based Models",
author = "Felzenszwalb, P. F. and Girshick, R. B. and McAllester, D. and Ramanan, D.",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
year = "2010", volume = "32", number = "9", pages = "1627--1645"}