ygh96521 / gan-visual-tracking

Generative Adversarial Networks for Visual Object Tracking

Generative Adversarial Networks for Online Visual Object Tracking Systems

Introduction:

This project is to build visual object tracking benchmark using generative adversarial networks to track any moving target in a sequence without prior traning on the target.

This implemntation is based on MDNET, VITAL and RT-MDNet.

I modified the architecture of the baseline to get MDGanet, ROIAL-MDNet, MDResNet and MDResGaNet trackers.

For more information:

• Thesis

• Presentation

• Demo

Usage

For Tracking

python run_trackers.py  -t <trackers> -s <sequences> -e <evaltypes> -n <testname> 

For example to run ROIAL tracker:

python run_trackers.py  -t RoialMDNet -s Basketball -e OPE -n tb50 

For Pre-Training

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

• Pretraining on VOT-OTB:

  • Download VOT datasets into "datasets/VOT/vot201x"

• Pretraining on ImageNet-VID

  • Download ImageNet-VID dataset into "datasets/ILSVRC"