Add circle model to the RANSAC gallery example #5208
Conversation
|
Hello @sciunto! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:
Comment last updated at 2021-01-29 09:29:19 UTC |
| return x, y | ||
|
|
||
|
|
||
| # Generate three types of data points |
There was a problem hiding this comment.
Can you briefly describer what the purpose of these 3 are and which one we are expecting to estimate with the circle model?
| xy = np.stack([x, y], axis=1).astype(float) | ||
|
|
||
| # Fit with RANSAC | ||
| ransac_kwarg = {'min_samples': 100, |
There was a problem hiding this comment.
I am not familiar with RANSAC, but looking at the output qualitatively I was wondering why only about half of the circle shows up as inliers in the right panel of the plot?
empirically it seems that changing min_samples to a value <= 80 helps get many more inliers along the circle boundary.
| ax[0].plot(x2, y2, '.') | ||
| ax[0].plot(x1, y1, '.') | ||
|
|
||
| circle = plt.Circle((cy, cx), radius=r, alpha=0.1) |
There was a problem hiding this comment.
increase alpha to 0.3 or so? I didn't really even see the disk at first!
| ax[1].add_patch(circle) | ||
| ax[1].plot(xy[inliers, 0], xy[inliers, 1], | ||
| 'bo', markersize=1, label='inliers') | ||
| ax[1].plot(xy[~inliers, 0], xy[~inliers, 1], |
There was a problem hiding this comment.
outliers was defined above. Did you intend to use that here instead of ~inliers?
grlee77
added
the
📄 type: Documentation
Description
To investigate #5186 I wrote this example that I propose to add to the gallery.
Checklist
./doc/examples(new features only)./benchmarks, if your changes aren't covered by anexisting benchmark
For reviewers
later.
__init__.py.doc/release/release_dev.rst.