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
make precision_recall_curve plots less confusing #7372
Comments
What's wrong with Pascal VOC might be more complicated. Maybe we could add dotted vertical lines indicating each recall threshold, and circle the operating points that are selected for averaging. |
Nothing is wrong with this. I think this is the right way. Maybe some shading to illustrate the area? So I think there is two things here: fixing the existing examples with |
@amueller I've now pushed an update to my branch for this example. I've also written a minimal example that highlights the difference between the old and new implementations, but I'm not sure where that would go. For now, here it is:
This produces the following plot: |
I want to make my first contribution . Can I work on this? |
I am closing this issue because the |
Currently the "precision_recall_curve" plots in the examples are kind of confusing because they use linear interpolation, which is pretty bad.
We could use
plt.step
instead ofplt.plot
but that would look a bit weird. We could also interpolate the curve on a fixed grid to get a Pascal VOC like thing.In addition, I think we should have an explanation of the different ways average precision and the precision recall curve are calculated.
I might work on that tomorrow.
The text was updated successfully, but these errors were encountered: