Implementation of RandAugment in 3D
Project description
RandAugment-3D
Implementation of RandAugment in 3D
Original paper: arXiv:1909.13719
Original implementations (in tensorflow):
- From official paper: https://github.com/tensorflow/tpu/blob/298d1fa98638f302ab9df34d9d26bbded7220e8b/models/official/efficientnet/autoaugment.py
- More recent: https://github.com/tensorflow/models/blob/1c79ece9f43340e9bc9571e06a4bf9bd8db8d97a/official/vision/beta/ops/augment.py
Most functions used here are implemented by MONAI.
Installation
pip install RandAugment3d
How to use
Args:
- n: number of augmentations to apply
- magnitude: magnitude of augmentations to apply, on a scale where 10 is full level (this scale is used to be as similar as possible to the original implementation). Values > 10 result in more distortion, values < 10 in less.
- excluded_operations: list of names of the excluded operations. Valid names are listed above.
operations names: ['identity', 'rotate_x', 'rotate_y', 'rotate_z', 'translate', 'scale', 'shear_x', 'shear_y', 'shear_z', 'shiftIntensity', 'equalize', 'solarize', 'histogramShift', 'sharpen', 'adjustContrast']
augment = RandAugment3D(n=2, magnitude=10)
Geometric Augmentations only
augment = RandAugment3D(n=2, magnitude=10, excluded_operations=['shiftIntensity',
'equalize',
'solarize',
'histogramShift',
'sharpen',
'adjustContrast'])
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Close
Hashes for RandAugment3d-0.0.3.dev0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | a49d7d343f88f5c5ec697cf8fd0aa1933fb4ce4c07b1f627eaef4b2df3051649 |
|
MD5 | 6864db2ffb4d228367dab53141d885e8 |
|
BLAKE2b-256 | 2f32707ef7569c8a8c582b1a63d9ea5d1b35c9aa6961af65aaeb35f452d526c6 |