Skip to content
Provides users a possibility to train their own GOTURN tracker CNN model on custom data
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.gitattributes
.gitignore
README.md Update README.md Aug 21, 2016
buildH5Dataset.h
buildH5Datasets.cpp
goturnDeploy.prototxt
goturnSolver.prototxt
goturnTrain.prototxt
main.cpp
testDatasets.txt
trainDatasets.txt
utils.cpp
utils.h

README.md

GOTURN Training Toolkit

This is the code for training of GOTURN tracker implemented inside OpenCV.

Original GOTURN 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:

sudo apt-get install libopencv-dev

Building HDF5 dataset

For GOTURN training, first HDF5 dataset shold be generated. It can be done by simple function call:

buildDB()

By default it generates 10 HDF5 datasets every with 500x10 samples (10 crop samples per image as proposed by authors). As alternative more low-level function can be used:

buildH5Datasets(datasetName, numberOfSamples)

Training network

GOTURN training requires a "bvlc_reference_caffenet.caffemodel" for GOTURN network weights inialization:

http://dl.caffe.berkeleyvision.org/bvlc_reference_caffenet.caffemodel

Training is launched by next line:

trainNet()

All hyperparameters are configured in goturnSolver.prototxt, for more details refer Caffe documentation.

Evaluate the tracker

In order to visualize results, GOTURN tracker can be tested on test dataset:

buildH5Datasets("D:/ALOV300++/trainDataset.h5", 10);
testNet("goturn_iter_30000.caffemodel");

First command generate a new small test dataset, and second launching a test procedure with visualization.

Pretrained GOTURN model

Also there is pretrained GOTURN model is available in OpenCV_extra repository

https://github.com/opencv/opencv_extra

You can’t perform that action at this time.