search

martin-danelljan / ECO

Matlab implementation of the ECO tracker.
GNU General Public License v3.0
612 stars 248 forks source link

About sample weights and scale estimation #2

Closed gzpan closed 7 years ago

gzpan commented 7 years ago

I would like to know why CCOT and ECO do not use adaptive decontamination strategy as SRDCF_decon do and scale estimation as fDSST do? Isn't SRDCF_decon an effective method to solve the problem of model update? Though scale estimation strategy from fDSST is included in ECO, it is not used in OTB and VOT test. I am confusing.

martin-danelljan commented 7 years ago

When we develop our trackers, we prefer to keep things simple and transparent. Although the decontamination approach, which is quite general, can give a boost in performance, it was not necessary for CCOT and ECO in order to demonstrate their effectiveness. The decontamination CVPR 2016 paper is an independent contribution that can easily be integrated in either of these two. In fact, the code in CCOT and ECO are already prepared for that.

gzpan commented 7 years ago

Oh I see. thanks! I want to say that you always do the right thing in tracking. Hope keeps it and contribute more elegant ideas to tracking(CV) community.

martin-danelljan commented 7 years ago

Thank you for your kind words! :D