fastlbp-imbg 0.0.1

Parallel multiradial LBP features

fastLBP

Highly parallel LBP implementation

Installation

Підготовка середовища пітона (краще робити не на head node)

• Знайти десь Python 3.11 (напр. conda create -n p11 python=3.11)
• Перевірити що середовище правильне
  • python --version та pip --version
• Встановити пакет через
  • pip install fastlbp-imbg
• Або вручну через
  • git clone [e-mail]:imbg-ua/fastLBP.git
  • pip install fastLBP

Наша директорія на lustre /lustre/scratch126/casm/team268im/

Implemented modules

run_skimage

Computes multiradial LBP of a single multichannel image in a parallel fashion.
This is a quick sample implementation that could be a baseline for further benchmarking.

Features:

• Powered by skimage.feature.local_binary_pattern
• Concurrency is managed by Python's multiprocessing module
• Parallel computation via multiprocessing.Pool of size ncpus
• Efficient memory usage via multiprocessing.shared_memory to make sure processes do not create redundant copies of data
• It computes everything in RAM, no filesystem usage

TODO:

• Use max_ram parameter to estimate optimal number of sub-processes and collect memory stats. Now max_ram is ignored.

Planned modules

run_chunked_skimage

Similar to 1. run_skimage, but each subprocess should compute LBP for its image chunk, not the whole image.

run_dask and run_chunked_dask

Similar to 1. run_skimage, but use Dask and dask.array.map_overlap for parallelisation instead of multiprocessing and manual data wrangling

Other notable things to try

• Perform benchmarking of an in-house optimised cython version of skimage.feature.local_binary_pattern (see skimage_lbp.pyx at imbg-ua/nf-img-benchmarking)
• Do some research on Numba - is it applicable here?
• Add annotations to run_skimage results using anndata