Wzxthu / py-MDNet-Filter-Evolution

MDNet python implementation

py-MDNet

by Hyeonseob Nam and Bohyung Han at POSTECH

Introduction

Python (PyTorch) implementation of MDNet tracker, which is ~2x faster than the original matlab implementation.

[Project] [Paper] [Matlab code]

If you're using this code for your research, please cite:

@InProceedings{nam2016mdnet,
author = {Nam, Hyeonseob and Han, Bohyung},
title = {Learning Multi-Domain Convolutional Neural Networks for Visual Tracking},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2016}
}

Prerequisites

• python 2.7
• PyTorch and its dependencies

Usage

Tracking

 cd tracking
 python otb_demo.py -s DragonBaby [-d (display fig)] [-f (save fig)]

• You can provide a sequence configuration in two ways (see tracking/gen_config.py):
  • python run_tracker.py -s [seq name]
  • python run_tracker.py -j [json path]

Pretraining

• Download VGG-M (matconvnet model) and save as "models/imagenet-vgg-m.mat"
• Download VOT datasets into "dataset/vot201x"

 cd pretrain
 python prepro_data.py
 python train_mdnet.py

Test with VOT-Toolkit

• Download the VOT-Toolkit following the instructions.
• Create a workspace named "mdnet".
• Copy the tracker_mdnet.m to the workspace to replace the generated one.
• If you are using PyTorch 0.4.x, please downgrade it to 0.3.1, as 0.4.x has compatibility problem with Matlab.