Image augumentation for deep learning.
Renames all image files to SHA-1 format (sha1.jpeg/png)
./pyImageAug.py -i test/
Renames all image files to SHA-1 format (sha1.jpeg/png) and creates new augmented files based on original files with random transformations like
- gauss noise added
- rotatation
- translation
- perspective transformation
- blur
- flip horizontally
- affine transformation
- hue
- saturation
- contrast
- brightness
- mirror and mosaic
- mosaic from 4 images
- night
- vignette
- rain
- blackboxing
and mixes of all these methods.
./pyImageAug.py -i test/ -a
usage: pyImageAug.py [-h] [-i INPUT] [-ar] [-v]
optional arguments:
-h, --help show this help message and exit
-i INPUT, --input INPUT Input path
-ar, --augmentation Process extra image augmentation.
-v, --verbose Show verbose finded and processed data