Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

What exactly do templates refer to in template matching #4

Closed
longmalongma opened this issue Apr 29, 2021 · 1 comment
Closed

What exactly do templates refer to in template matching #4

longmalongma opened this issue Apr 29, 2021 · 1 comment

Comments

@longmalongma
Copy link

image

I would like to ask the following questions: 1. What exactly do templates refer to in template matching; 2. Refine-translator is added after backbone?

@HYOJINPARK
Copy link
Owner

HYOJINPARK commented May 2, 2021

Hi @longmalongma
Thanks for asking.

diagram

The 'template' has target information and embeds A, C region as a dissimilarity map (D_t).
Actually, I did not mention this detailed motivation in my paper....
I conjecture that:

When a target moves from red to blue region, the S_(t-n) gives attention to the red region (A+B).
Therefore, we know where is A + B region in f8_t
However, the real target information is located in B + C region inside of f8_t.
Therefore, B has strong target information (attention from S_t-n and target feature),
and C (target feature) has less strong target information.
The template matching uses the 'template' and recognizes the B and C region from f8_t.
So finally, we can know and divide where is B and C regions,
and creates a dissimilarity map (D_t) for A and C region.

Yes, refine-translator is added after backbone (feature extractor)
You can insert refine-translator middle of generating mask (refined network).
Please let me know if you have further questions.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants