impart 0.0.5

Partition images and volumes of arbitrary size

impart (image partitioning) provides a simple representation of images and volumes (of arbitrary size) as a partition (i.e., grid) of fixed-size cells, with a given step size between cells.

Partitions can be indexed and sliced like NumPy arrays. Any extracted image region is only loaded into memory after calling .numpy(). Complex indexing logic is abstracted. impart also includes some utilities, e.g., multi-core fragmentation of partitions into their individual cells.

Viewing image sub-regions as cells with a fixed step is useful for data processing in several domains, such as remote sensing, medical imaging, and deep learning.

Installation, usage & benchmarks

Please refer to the official documentation:

https://impart.readthedocs.io/en/latest/