twotoone / concise-goturn

concise implementation of GOTURN,Source code for paper: Learning to Track at 100 FPS with Deep Regression Networks

Concise GOTURN

This is a concise implementation for GOTURN: Generic Object Tracking Using Regression Networks.

This project is based on project GOTURN, I just remake GOTURN and delete some redundant files so that I can test my video Conveniently.

I appreciate that David Held opened his code and trained caffemodel. According to his code and help document, we can easily re-run this project and test this algorithm. But this complete project GOTURN have some redundant files, which will add difficults for our testers,for example folder "train","test" and "visualizer". So I implement this concise project, so that we can test this algorithm by our own video more easily.

GOTURN appeared in this paper:

Learning to Track at 100 FPS with Deep Regression Networks,
David Held, Sebastian Thrun, Silvio Savarese,
European Conference on Computer Vision (ECCV), 2016 (In press)

Installation

Install dependencies:

• Install CMake:

sudo apt-get install cmake

• Install Caffe and compile using the CMake build instructions: http://caffe.berkeleyvision.org/installation.html You must compile caffe by cmake. If you do not use cmake to compile caffe , this project can not find required caffe.

• Install OpenCV

sudo apt-get install libopencv-dev

If you installed opencv, do not execute it.

Compile

From the main directory, type:

open CMakeLists.txt,and change set(Caffe_DIR your_caffe_folder),for example, mine is set(Caffe_DIR ~/tracking/GOTURN/caffe)

then

mkdir build
cd build
cmake ..
make

Pretrained model

You can download a pretrained tracker model (434 MB) by running the following script from the main directory:

bash scripts/download_trained_model.sh

Test your own video

bash scripts/runTracker.sh