Oguzhan Gencoglu

Morphology benchmarks to torch (CPU), for example kornia

4

kornia has [torch-native morphology ops](https://kornia.readthedocs.io/en/latest/morphology.html). Benchmarks to that would be great. I guess accounting numpy to torch and back to numpy conversions to have apples to apples. All on CPU...

pip installation error on Mac

2

Here is the full log of `pip install imops` on Mac, Python3.11, virtualenv: ``` Building wheels for collected packages: imops Building wheel for imops (pyproject.toml) ... error error: subprocess-exited-with-error ×...

`from string2string.metrics import sacreBLEU` raises the error: ``` File ~/my_dir/lib/python3.9/site-packages/string2string/metrics/rouge.py:8 1 """ 2 This module contains a wrapper class for the ROUGE metric. 3 4 ROUGE (Recall-Oriented Understudy for Gisting...

It is not clear if this is only US or does it also have other countries? If so, which? Would be nice to get a rough understanding without having to...

Inference on arbitrary images

1

Thanks for the great work! Any Colab demo on inference on arbitrary images (not in ImageNet)?

What exactly is meant by "temporary model" in README?

Thanks for the great work. Would you be able to provide an inference script that works with a single example input graph so that there is no need to fill...

Thanks for the interesting work. Is there any plans to be able to support exporting the rules in ONNX format (`sklearn-onnx` supports decision trees at least) for ease of deployment?

Huggingface demo link in docs does not allow user uploaded images

2

## Describe the bug Huggingface demo link in docs does not allow user uploaded images. ### To Reproduce Test here: https://huggingface.co/spaces/qubvel-hf/albumentations-demo?transform=CLAHE

Correct way of converting HistGradientBoostingRegressor with categorical_features?

2

HistGradientBoostingRegressor supports specification of `categorical_features`: https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.HistGradientBoostingRegressor.html#sklearn.ensemble.HistGradientBoostingRegressor What would be the correct `initial_type` in `convert_sklearn(model, initial_types)` to achieve this behaviour. Float inputs to ONNX are giving wrong predictions of course.