Skip to content
Code of BEVT-Tracker v1.0 (Matlab Version for Discussion)
C MATLAB C++
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.
external Tracker (v1.0) and data sequences Feb 28, 2019
model Tracker (v1.0) and data sequences Feb 28, 2019
response Tracker (v1.0) and data sequences Feb 28, 2019
result add results to readme Jul 10, 2019
seq Tracker (v1.0) and data sequences Feb 28, 2019
util Resolved some compatibility issues Jul 30, 2019
.DS_Store Tracker (v1.0) and data sequences Feb 28, 2019
.gitignore
BEVT_Demo_single_seq.m Resolved some compatibility issues Jul 30, 2019
BEVT_tracker.m Tracker (v1.0) and data sequences Feb 28, 2019
README.md
configSeqs_demo_for_BEVT.m
run_BEVT.m

README.md

BEVT-tracker

Test passed
matlab-2017b MatConvNet-1.0--beta25 CUDA-8.0

Matlab implementation of Boundary Effect-Aware Visual Tracking for UAV with Online Enhanced Background Learning and Multi-Frame Consensus Verification (BEVT tracker).

Instructions

  1. Download VGG-Net-19 by cliking here and put it in /model.
  2. Download matconvnet toolbox here and put it in /external.
  3. Configure the data sequence in configSeqs_demo_for_BEVT.m.
  4. Run BEVT_Demo_single_seq.m

Note: the original version is using CPU to run the whole program. If GPU version is required, just change false in the following lines in run_BEVT.m to true:

global enableGPU;
enableGPU = false;

vl_setupnn();
vl_compilenn('enableGpu', false);

Acknowledgements

The parameter settings are partly borrowed from BACF and SRDCF paper and convolutional feature extraction function is borrowed from HCF.

Results

The following are the results from the experiment conducted on 100 challenging sequences extracted from UAV123@10fps.

You can’t perform that action at this time.