Skip to content

Updated deprecated functions and types - #123

Open
TheoFABIEN wants to merge 4 commits into
dyhBUPT:masterfrom
TheoFABIEN:master
Open

TheoFABIEN wants to merge 4 commits into
dyhBUPT:masterfrom
TheoFABIEN:master

Conversation

@TheoFABIEN

Copy link
Copy Markdown

This pull request updates deprecated functions and data types to ensure compatibility with recent versions of the dependencies, as the repository does not specify package versions.

Changes include:

  • Replaced sklearn.utils.linear_assignment_ with scipy.optimize.linear_sum_assignment (since linear_assignment is deprecated in sklearn).
  • Replaced np.int with int (deprecated in NumPy).
  • Replaced np.float with float (deprecated in NumPy).

Replaced the deprecated linear_assignment from sklearn with the linear_sum_assignment from scipy + slight adaptations to the output format of this new function.
np.int is deprecated in recent version, so I replaced it with the built-in int.
Replaced the deprecated np.int with the built-in int.
np.float is deprecated alias for the built-in float, so I replaced it with float.
@dyhBUPT

dyhBUPT commented Feb 25, 2025

Copy link
Copy Markdown
Owner

Hi, thanks for your PR!
After your changes, are the evaluation results consistent with our reported results in paper?

@TheoFABIEN

Copy link
Copy Markdown
Author

With the updated code I was able to successfully run object tracking on the MOT16 dataset.

Then I did some unit testing to double check that my modification for linear assignment gives strictly identical results to the old one.

np.int and np.float were aliases for python types int and float, respectively, and numpy documentation advises to use int and float as replacements.

Unfortunately I did not have enough time to run the evaluation results to compare directly with the paper, but I am confident that they will be the same, given the packages documentation and the tests I did.

If I find some time to do that I could try to run the evaluation results, if you think it is necessary.

@dyhBUPT

dyhBUPT commented Feb 26, 2025

Copy link
Copy Markdown
Owner

Thanks for your explanation!
To ensure the consistent tracking performance, I will merge this PR after you check the evaluation results on MOT17-val set.

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

Successfully merging this pull request may close these issues.

2 participants