AugmentIt

This package gives basic augmentation techniques like cropping, resizing, scaling etc, for classification, segmentation and object detection.

Note:

This package is still in development.Now only it supports image cropping, image resizing with their annotations and yolo format conversion.

Installation

Install

pip install augmentit

Documentation

crop_image(image, coords, roi_coords, draw_image)

image - Numpy image array
coords - List of annotations [[[x1, y1], [x2,y2],..], ...]
roi_coords - coordinates to crop the image [[x1, y1], [x2,y2],..]
draw_image - True or False to verify the conversion

resize_image(image, coords, roi_coords, draw_image)

image - Numpy image array
coords - List of annotations [[[x1, y1], [x2,y2],..], ...]
roi_coords - coordinates to crop the image [[x1, y1], [x2,y2],..]
draw_image - True or False to verify the conversion

yolo_format(size, box)

size: size of the image [width, height]
box: coordinates [[xmin,ymin], [xmax, ymax]]

Returns: Converted yolo coordinates [[x,y], [w,h]]

plot_one_box(x, img, color=None, label=None, line_thickness=None)

x: [[xmin,ymin], [xmax, ymax]]
img: Image array

Return: Image array

plot_one_polygon(x, img, color=None, label=None, line_thickness=None)

x: [[[x1,y1], [x2, y2], ..], ..]
img: Image array

Return: Image array