tile_and_stitch

Clone the repo

https://github.com/danrustia11/tile_and_stitch

Sample usage

import tools.image_tiler as tl
import cv2

# Define a tiler object

tiler = tl.tiler(TILING_SIZE, PADDING)

# Open an image

image = cv2.imread([filename])

# Remove the borders from the image to make it fit for tiling

no_border_image = tiler.remove_borders(image)

# Apply tiling

tiled_images = tiler.tile_image(no_border_image)

Preparing data using the library

Prepares images with cut borders (fit for tiling) and tiled images. Labels are also retranslated based on the tiled output images.

Script name:
1_data_preparation.py

Arg	Input	Description
--dir	string	Source image directory
--split	string	Percentage of train, validation, and test
--tile	int	Tiliing size
--padding	int	Padding size (overlap)
--verbose	int	To print system messages or not (0 or 1)
--density	string	Divides the images into low, medium, and high counts; ignored if no input is provided

If without labels:

python 1_data_preparation.py --dir [directory_name] --tile 1200 --padding 200

If with labels:

[With density splitting]
python 1_data_preparation.py --dir [directory_name] --tile 1200 --padding 200 --density 10,30

[Without density splitting]
python 1_data_preparation.py --dir [directory_name] --tile 1200 --padding 200