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

Only save results with a given confidence percentage or above #496

Open
1 task done
JaimyvS opened this issue May 8, 2024 · 2 comments
Open
1 task done

Only save results with a given confidence percentage or above #496

JaimyvS opened this issue May 8, 2024 · 2 comments
Assignees
Labels
enhancement New feature or request question Further information is requested

Comments

@JaimyvS
Copy link

JaimyvS commented May 8, 2024

Search before asking

  • I have searched the Pytorch-Wildlife issues and found no similar bug report.

Question

Is there currently a function in Pytorch Wildlife to only export results when they are above a given confidence level. Either for the JSON output or the cropped output. If not, I would love to put this in as a feature request.

Thanks!

Additional

No response

@JaimyvS JaimyvS added the question Further information is requested label May 8, 2024
@zhmiao
Copy link
Collaborator

zhmiao commented May 8, 2024

Hello @JaimyvS, currently we don't have a function that filter the outputs by confidence. It will be a good postprocessing function in addition to the current ones. We will be happy to add it in our future iterations!

@zhmiao zhmiao added the enhancement New feature or request label May 8, 2024
@Wologman
Copy link

I would have prefered the original behaviour for the .json output, though I could see why others might want them filtered. Perhaps we could have this as an optional flag rather than a built-in behaviour?

For my purposes, we are still interested performing further analysis on the empty images. For example to calibrate our models against human verified examples, or to make decisions about appropriate thresholds. If the empty images (or images where the confidences are below the threshold) are not included in the .json files, then I need to handle it myself by adding them back in, which is OK, but extra complexity.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request question Further information is requested
Projects
None yet
Development

No branches or pull requests

4 participants