Skip to main content

Playground ML Unified Microlib Set: The Playground ML python toolbox package

Project description


PLaygroundML Unified Microlib Set : The Playground ML python toolbox package

pyversions License

The Plums library set aims to defined a common set of packages to be used by people involved in thePlaygroundML team.

Those packages puropose is to set a unique baseline to help make the code base more unified and avoid countless reimplementation of the same tools which in turns make people waste time and make the code base herd to understand, debug and reuse.

Installation is simple with PyPI repository:

pip install plums

[TODO] More information on installation can be found in the Getting Started section of the documentation.

Documentation and tests specific dependencies can be installed with the docs and tests extra keywords respectively.



The Plums commons package aims to offer a set of lightweight highly reusable classes and utilities for all other packages.

To import do:

import plums.commons


The Plums plot package aims to offer a set of lightweight highly reusable classes and utilities for visualizing detection and segmentation results.

To import do:

import plums.plot


The Plums model package aims to offer a framework-agnostic model format specification (the Plums Model Format) along with its python representation and helper implementation to ease integration into producer and consumer codebases.

To import do:

import plums.model


I/O operations and efficient dataset iteration and indexing handling.

To import do:

import plums.dataflow



Dataflow elements to speedup future developements: e.g. Dataset classes, Sampler and/or Dataloader (?) and handle augmentation (imgaug, albumentation, pure numpy ?)

  • Python representation of data elements (e.g. Annotation, Feature, Image, Datapoint...)
  • Dataset classes for playground datasets
  • TranformedDataset-like classes to manipulate, combine and transform (e.g. augmentations) datasets in an online fashion
  • Sampler and BatchSampler classes to port PyTorch functionalities into Keras and build upon them


  • Loop through datasets in a multi-threaded or multi-processed fashion
  • Some convenient data-preparation functions such as image transformations or annotations refinement
  • Some convenient statistic and analysis tools


  • Plot annotation on image
  • Handle single image and image grids
  • Plot differentials and/or superposition of annotations on same images
  • Handle multiple color code mode, e.g.:
    • By label
    • By confidence
    • By size
    • By type (in differential plotting)
    • And possibly combination of examples above (e.g. Color by label and shades by confidence)

Model Format (UMF like ?)

  • Python representation of a model and its format components
  • IO functionalities, e.g.:
    • Save a model (as collect disparate resources into a coherent model directory w/ metadata)
    • Load a model (as create a Python model representation w/ metadata from a model directory)
    • Verify an existing model directory
    • Copy and/or prune a model

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

plums-0.5.2.tar.gz (286.2 kB view hashes)

Uploaded source

Built Distribution

plums-0.5.2-py2.py3-none-any.whl (313.0 kB view hashes)

Uploaded py2 py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page