Add unet encoder-decoder and image output feature

[vijayi1]

Code Pull Requests

Please provide the following:

• a clear explanation of what your code does
  Added a UNet encoder and decoder stack.
  Added an image output feature, and image utils to detect pixel classes.
  Added an image semantic segmentation example using a kaggle dataset.
  Added test cases for encoder, decoder and dataset.
• if applicable, a reference to an issue
• a reproducible test for your PR (code, config and data sample)
  Test example included in PR.

[vijayi1]

@justinxzhao, I opened this PR and closed the previous PR 3870 due to git rebase issues. I've addressed your comments in this PR.

I'm attaching some sample results. Each result has the input image, the label mask and a predicted mask.

1. the camseq example included in the PR (https://www.kaggle.com/datasets/carlolepelaars/camseq-semantic-segmentation)
2. a buildings dataset, with images cropped due to mem limits on my system (https://www.kaggle.com/datasets/sohaibmoradi/segmented-satellite-images-of-buildings)

[github-actions]

Unit Test Results

  6 files  ±0    6 suites  ±0   13m 50s ⏱️ -26s
12 tests ±0    7 ✔️  -   2    5 💤 +  2  0 ❌ ±0 
60 runs  ±0  30 ✔️  - 12  30 💤 +12  0 ❌ ±0 

Results for commit 30feafb. ± Comparison against base commit 9bb89c6.

This pull request skips 2 tests.

tests.regression_tests.benchmark.test_model_performance ‑ test_performance[ames_housing.gbm.yaml]
tests.regression_tests.benchmark.test_model_performance ‑ test_performance[mercedes_benz_greener.gbm.yaml]

[vijayi1]

attached training results for the camseq example above.

[justinxzhao]

@vijayi1 This is super cool! Thanks for addressing prior comments and putting together the PR. LGTM!

[arnavgarg1]

PR looks great!! Amazing work!

Non-blocking, but two quick follows up (can be done after this PR is merged):

1. Would you be able to add unit tests for the new functions added in image_utils.py? Also happy for you to create a follow up PR to add tests for those util functions!
2. Are you also able to create a PR in https://github.com/ludwig-ai/ludwig-docs for the new image encoder and decoder? We can do it after we merge this PR again,

[vijayi1]

Thanks for the reviews. @arnavgarg1, I'll create follow up PRs for those two points.

I do not have write access to merge this PR. Please advise on the next step.

[arnavgarg1]

@vijayi1 - Sounds good! Will keep a lookout for those PRs. We just merged your PR into master!

[arnavgarg1]

Hey @vijayi1! This was a marquee feature in our Ludwig 0.10.0 release and we'd love to give you a shoutout on LinkedIn. Is there a way to connect? My LinkedIn is https://www.linkedin.com/in/arnavgrg/, feel free to connect!

[vijayi1]

Hi @arnavgarg1, thanks for the shoutout. I'm not on LinkedIn, but I did lookup the release announcement.

[arnavgarg1]

No worries! And glad you were able to see it :) Just wanted to make sure you knew and we really appreciate the contribution!

[alexsherstinsky]

@vijayi1 What is your handle in the Ludwig community Slack? Thank you!

[vijayi1]

"Vijay I", I just joined.