Fix #545: Add "path" arg and deprecate "raster_path" in predict_tile()

[Samia35-2973]

I sincerely apologize for the inconvenience I caused earlier. I truly appreciate your patience and guidance. If it’s okay, I have now tried to organize everything freshly with an updated version.

This PR solves the issue #545 by updating the predict_tile function to use path as an alternative to raster_path while issuing a deprecation warning when raster_path is used. I have also tried to maintain the test cases.

Here's what I modified-
In main.py:

• Introduced path as a replacement for raster_path.
• Issued a DeprecationWarning when raster_path is provided.
• Updated all references to raster_path to use path.

In test_main.py: Updated the test cases to use path instead of raster_path.

In test_visualize.py: Modified test_predict_tile_and_plot to use path instead of raster_path.

I did not modify the following function in test_dataset.py

def raster_path():
    return get_data(path='OSBS_029.tif')

Please review this and let me know if this function should also be updated along with any further modifications.

[henrykironde]

Ref Same: #975

[Samia35-2973]

Sorry again for the trouble. Due to my lack of experience, I mistakenly force-pushed my branch, which unintentionally closed my previous PR. Since I wasn’t sure how to proceed, I submitted a new PR, but I now realize that might not have been the best approach. Could you kindly guide me on what the best course of action is from here?

[Abhishek-Dimri]

Hi @Samia35-2973, no worries—this happens! Instead of creating a new PR, you can update your existing one like this:

1. Make your changes.
2. Amend the commit:

   git add <file>  
   git commit --amend  
   git push -f origin your-branch  

This updates the PR without adding new commits.

I faced a similar issue before, and mentor helped me to resolve it. You can see how they guided me here: PR #930.

Hope this helps! 😊

[Samia35-2973]

Thanks @Abhishek-Dimri, that makes a lot of sense. Now that I opened this new PR, what do you think I should do? Should I close this new PR and try to recover the old one?

[Abhishek-Dimri]

@Samia35-2973, it’s your choice! You can reopen the old PR and update it, or you can continue working on this new PR. I don’t think either will cause any major issues.

Could you share why you made these changes and how they’ll help improve the project?

[Samia35-2973]

The changes I made are meant to align the argument names in predict_tile() and predict_image() as addressed in this issue #545. Right now, predict_tile() uses raster_path, while predict_image() uses path in the main.py file. The issue mentioned that there is no need to have more than one name for the same thing, so I’ve made predict_tile() use path instead and issued a deprecation warning for raster_path along with modifying the test cases.

[henrykironde]

Thanks @Samia35-2973 and @Abhishek-Dimri for the contribution.

[henrykironde]

These updates look good so far. We can add a few more things.

Let's search the entire repo for predict_tile as shown here and review all instances to ensure full coverage. This will enable you update the documentation accordingly.

For these changes, please make a new commit.

[Samia35-2973]

Hi @henrykironde, I have some questions regarding the changes.

1. In docs/user_guide/examples/nest_detection.ipynb, there's a code snippet:

   # Add a path to an image to test the model on
   raster_path = ""
   predicted_raster = model.predict_tile(raster_path,
                                         return_plot=True,
                                         patch_size=300,
                                         patch_overlap=0.25)
   plt.imshow(predicted_raster)
   plt.show()

   I will change the raster_path to path for that cell. Do you want me to rename predicted_raster to predicted_image, or should I keep it as is?

2. I am only adding the deprecated argument of predict_tile() in history.rst. Do I need to add anything else, like updating the use of path instead of raster_path in any versions?

Thanks!

[henrykironde]

I will change the raster_path to path for that cell. Do you want me to rename predicted_raster to predicted_image, or should I keep it as is?

You can edit that code as you feel. Make sure the readers can understand the code easily.

2. I am only adding the deprecated argument of predict_tile() in history.rst. Do I need to add anything else, like updating the use of path instead of raster_path in any versions?

To update HISTORY.rst, copy the previous template and add the new entry. I periodically collect and clean up past commits for a summary. Since we don’t have a version title yet, use a placeholder like Version x.x.x (Date: ). Feel free to include any major changes we have merged in like that of paths.

[Samia35-2973]

Hi @henrykironde, I have updated all occurrences of raster_path to path argument of predict_tile() in the documentation (tutorials, user guides, examples).
I also added the deprecation notice in HISTORY.rst as discussed.
I'm happy to make any additional changes needed to ensure full coverage of this update.

[henrykironde]

Thanks @Samia35-2973 for those changes