py-distance-transforms 0.2.0

Add your description here

py-distance-transforms

py_distance_transforms is a Python package that provides efficient distance transform operations on arrays. It is a wrapper around the Julia package DistanceTransforms.jl, bringing its high-performance capabilities to the Python ecosystem.

Documentation

Docs	Description
Getting Started:	A quickstart guide to using py_distance_transforms for efficient distance transform operations on arrays.
Deep Learning (Hausdorff Loss):	A MONAI tutorial adjusted to show how to use the Hausdorff loss function and the corresponding py_distance_transforms

Features

• Fast distance transform computations on CPU and GPU
• Support for 1D, 2D, and 3D arrays
• Multi-threading for enhanced CPU performance
• GPU acceleration for NVIDIA GPUs (CUDA)
• Simple and intuitive API

Installation

Install py_distance_transforms using pip:

pip install py_distance_transforms

Basic Usage

from py_distance_transforms import transform_2d
import numpy as np

arr = np.random.choice([0, 1], size=(10, 10)).astype(np.float32)
result = transform_2d(arr)

GPU Acceleration

import torch
from py_distance_transforms import transform_gpu_2d

x_gpu = torch.rand((100, 100), device='cuda')
x_gpu = (x_gpu > 0.5).float()

gpu_transformed = transform_gpu_2d(x_gpu)

Acknowledgments

py_distance_transforms is a Python wrapper around the Julia package DistanceTransforms.jl.

• Huge thanks to @pabloferz for getting DLPack.jl to work with PythonCall/juliacall and PyTorch. Massive thanks to @cjdoris and all of the contributors to PythonCall.jl as well.