smh-data-augmentation 0.0.2

A data augmentator package

Data Augmentator

Description

This tool allows to generate modified data while keeping the statistical invariant, also known as data augmentation. It is useful to train machine learning models that need a huge amount of data and reduce the variance in predictions, see the Bias–variance tradeoff.

Features

Images

• Calculate new point positions.
• Recalculate bounding boxes.
• Command Line Interface

python3 -m data_augmentation.img -h

TODO

• PIP package
• Other domains than images.

Alternatives

This library is intented for internal uses of the lab so there could be others that fit better to your needs, see:

• imgaug