Skip to content
Code of KAOT-Tracker v1.0 (Matlab Version for Discussion)
Branch: master
Clone or download

Latest commit

Latest commit 2309322 Sep 20, 2019


Type Name Latest commit message Commit time
Failed to load latest commit information.
Version 1 add Version 2 May 20, 2019
Version 2 add Version 2 May 20, 2019 update Sep 20, 2019
performance_on_DTB70.png update Sep 20, 2019
performance_on_UAV123.png update Sep 20, 2019
precision.png add error and success fig May 20, 2019
success.png add error and success fig May 20, 2019

Keyfilter-Aware Real-Time UAV Object Tracking

Matlab implementation(Version 2.0)of Keyfilter-Aware UAV Object Tracker (KAOT).

Description and Instructions

How to run

  • download matconvnet
  • install.m - Compile libraries and download network
  • choose_KAOT_Demo.m - choose UAV123 image sequences for the KAOT tracker.


  1. Deep CNN features. It uses matconvnet [1], which is included in external_libs/matconvnet/. The imagenet-vgg-m-2048network available at was used. You can try other networks, by placing them in the feature_extraction/networks/ folder.
  2. HOG features.
  3. Colorspace features. Currently grayscale and RGB are implemented.

Performance on DTB70 and UAV123@10fps




This work borrowed the feature extraction modules from the STRCF tracker ( and the parameter settings from BACF (

You can’t perform that action at this time.