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

[Bug]: DLIB uses a small sliding window even when the upsampling limit is set to 0 #2825

Open
tintin-py opened this issue Jul 12, 2023 · 1 comment
Labels

Comments

@tintin-py
Copy link

tintin-py commented Jul 12, 2023

What Operating System(s) are you seeing this problem on?

Windows

dlib version

19.24.99

Python version

3.9.13

Compiler

GCC 6.3.0

Expected Behavior

I expect when the up-sampling limit is 0 then DLIB would use a sliding window that is approximately same as the bounding boxes that I have marked. I can't see any way of setting the sliding window size manually and this seems to be a major problem.

How is this sliding window dimension automatically picked up?

Current Behavior

The sliding window size is significantly smaller than the bounding boxes I have marked even when the up-sampling limit is set to 0. I can't set this window manually.

Steps to Reproduce

Here is the training.xml file

<?xml version='1.0' encoding='ISO-8859-1'?>
<?xml-stylesheet type='text/xsl' href='image_metadata_stylesheet.xsl'?>
<dataset>
<name>imglab dataset</name>
<comment>Created by imglab tool.</comment>
<images>
  <image file='patches\0.jpg' width='4962' height='3508'>
    <box top='180' left='625' width='2000' height='195'/>
    <box top='180' left='2590' width='2000' height='195'/>
  </image>
  <image file='patches\1.jpg' width='4963' height='3508'>
  </image>
  <image file='patches\4.jpg' width='4962' height='3508'>
    <box top='240' left='2856' width='2000' height='195'/>
    <box top='236' left='1232' width='2000' height='195'/>
  </image>
  <image file='patches\5.jpg' width='4963' height='3508'>
  </image>
  <image file='patches\8.jpg' width='4962' height='3508'>
    <box top='200' left='136' width='2000' height='195'/>
    <box top='196' left='1912' width='2000' height='195'/>
  </image>
  <image file='patches\9.jpg' width='4963' height='3508'>
    <box top='189' left='876' width='2000' height='195'/>
  </image>
</images>
</dataset>

As you can see, all the bounding boxes are of the same dimension. That's why I was expecting that up-sampling is not required and DLIB will just use the sliding window dimension same as the bounding boxes.

The training code is similar to that of the example provided.

Anything else?

No response

@tintin-py tintin-py added the bug label Jul 12, 2023
@tintin-py tintin-py changed the title [Bug]: [Bug]: DLIB uses a small sliding window even when the upsampling limit is set to 0 Jul 13, 2023
@Prathamesh01110
Copy link

Possible Reasons for the Issue:

1. Default Parameters: DLIB might have default parameters for sliding window sizes that are not automatically adjusted based on bounding box dimensions.

2. Configuration Issues: There might be a configuration or parameter setting that you are missing, preventing you from manually setting the sliding window size.

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

No branches or pull requests

2 participants