A plot_results function that works across predict* functions and multiple annotation types.

[bw4sz]

Quick roadmap for a number of visualization related issues.

• Create a single plot_results function that can take in points, boxes and polygons and plots them using supervision package, optionally saving a plot, and shows the image
• Confirm no backwards compatibility changes in plot_predictions
• Update docs to only use plot_results where possible.
• Make a decision on Consider decoupling plotting from predict_* functions #761 and close.

[bw4sz]

@ethanwhite this is ready for review. There is one backwards compatibility error. Within the predict_* functions, there is no longer a return_plot argument. This seemed cleaner than adding a deprecation.

[bw4sz]

Quick demo is within the doc build see docs/advanced_features/visuzalization.md in the readthedoc build

https://deepforest--768.org.readthedocs.build/en/768/advanced_features/visualizations.html

from deepforest import main, get_data
from deepforest.visualize import plot_results
import os

model = main.deepforest()
model.use_release()

sample_image_path = get_data("OSBS_029.png")
results = model.predict_image(path=sample_image_path)
plot_results(results, root_dir=os.path.dirname(sample_image_path))

[bw4sz]

Add root_dir as a class attribute of a prediction dataframe through pandas. Remove the root_dir argument from plot_results.

[bw4sz]

Make the visualizations larger. on the markdown document.

[ethanwhite]

Really excellent work @bw4sz and a lot of it. I've done my best to review the whole thing thoroughly. Almost all of the comments are minor cleanup, with the one slightly more major thing being a small change to how multi-class colors are handled. Let me know if you have any questions.

[ethanwhite]

@bw4sz - is this ready for me to give it another look?

[bw4sz]

yes, the error I see is not related.

…

On Thu, Sep 19, 2024 at 5:46 AM Ethan White ***@***.***> wrote: @bw4sz <https://github.com/bw4sz> - is this ready for me to give it another look? — Reply to this email directly, view it on GitHub <#768 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJHBLDWFJKLDIMO4EN4G3LZXLBSJAVCNFSM6AAAAABNVGTQMGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRQHA4TCNZUGI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Ben Weinstein, Ph.D. Research Scientist University of Florida

[bw4sz]

@ethanwhite I rebased and this should be ready.

[ethanwhite]

This all looks great. There was one remaining issue with our color handling in plot_results (my fault, I didn't have the fix quite right in my review). I just submitted a PR against this branch that I think fixes it.

@bw4sz - take a look at that PR and if everything looks good merge it and then we can get this merged.