/Coarse-to-fine-Tracker

Primary LanguageC++

Coarse-to-fine-Object-Tracking-using-Deep-Features-and-Correlation-Filters

This is the official code release for the paper : Zgaren A., Bouachir W., Ksantini R. (2020) Coarse-to-Fine Object Tracking Using Deep Features and Correlation Filters. In: Bebis G. et al. (eds) Advances in Visual Computing. ISVC 2020. Lecture Notes in Computer Science, vol 12509. Springer, Cham.

You can access it via arxiv:

For any questions or queries, please contact Ahmed Zgaren: ahmed.zgaren84@gmail.com

Prerequisites

  • Matlab R2018a
  • Neural Network Toolbox
  • VGG16 model
  • OpenCV

Dataset

The dataset used in the paper is from CVlab. It can be dowloaded from here

Getting Started

It is highly recommended to have a machine equipped with a GPU to speed up the tracking result. ( ~6 frame/s)

  • Download the dataset
  • Modify the file: configSeqs.m
    • on the OTB100 struct.
    • change the path for each sequence (add your local path where you stored the dataset).
  • Results (.mat) for each sequence are stored in the folder: results/results_OPE.
  • To execute the tracker, run the file: main_running.m.
  • To change the tracker parameters, access the file: tracker_benchmark v1.1\trackers\Ours_SRDCF\run_OURS_SRDCF.m