## Tracking with Kernelized Correlation Filters Code author : Tomas Vojir ________________ This is a C++ reimplementation of algorithm presented in "High-Speed Tracking with Kernelized Correlation Filters" paper. For more info and implementation in other languages visit the [autor's webpage!](http://home.isr.uc.pt/~henriques/circulant/). It is extended by a scale estimation (use several *7* different scales steps) and by a RGB (channels) and Color Names [2] features. Data for Color Names features were obtained from [SAMF tracker](https://github.com/ihpdep/samf). It is free for research use. If you find it useful or use it in your research, please acknowledge my git repository and cite the original paper [1]. The code depends on OpenCV 2.4+ library and is build via cmake toolchain. _________________ Quick start guide for linux: open terminal in the directory with the code $ mkdir build; cd build; cmake .. ; make This code compiles into binary **kcf_vot** ./kcf_vot - using VOT 2014 methodology (http://www.votchallenge.net/) - INPUT : expecting two files, images.txt (list of sequence images with absolute path) and region.txt with initial bounding box in the first frame in format "top_left_x, top_left_y, width, height" or four corner points listed clockwise starting from bottom left corner. - OUTPUT : output.txt containing the bounding boxes in the format "top_left_x, top_left_y, width, height" ./kcf_trax - using VOT 2014+ trax protocol (http://www.votchallenge.net/) - require [trax](https://github.com/votchallenge/trax) library to be compiled with opencv support and installed. See trax instruction for compiling and installing. ___________ Performance | | **VOT2016 - baseline EAO** | **VOT2016 - unsupervised EAO** | [**TV77**](http://cmp.felk.cvut.cz/~vojirtom/dataset/index.html) Avg. Recall | |:---------------|:--------------:|:------------------:|:----------------:| | kcf |0.1530 | 0.3859 |