PrepareImage - Custom Your Our Image Dataset

Flexible and Easy. We perform a package that help you build your own image dataset. Specially, it's easy to add your own prepare function

Install

pip instal prepareImage

How to use

1. Fork this repo and custom it!

2. Class dataset

Coming soon!

This is what make this package powerful.

3. Run directly on file prepare_data.py

1. Run on Terminal

If you already installed the package:

python3 -m prepareImage.prepare_data

or download file prepare_data.py from this repo and run:

python3 prepare_data.py

mode in ["split", "combine", "line", "pos", "all", "gray", "look_alike", "without_edge"]:

• combine: combine input, output images, then split to train/validation/test set
• line: add line to input images
• all: line then combine

Example:

python prepare_data.py --mode all \
--image_dir ../training_data/lhq_256 \ 
--caffemodel hed_pretrained_bsds.caffemodel \
--pos2_dir ../training_data/new_data/pos2 \
--pos 2 \
--grayscale_dir ../training_data/new_data/grayscale \
--line_dir ../training_data/new_data/edge \
--train_dir ../training_data/new_data/train \
--test_dir ../training_data/new_data/test \
--val_dir ../training_data/new_data/val \
--output_dir ../training_data/new_data/grayscale \
--combine_dir ../training_data/new_data/combine

Split Data:

python prepare_data.py --mode split \
--train_dir ../training_data/new_data/train \
--test_dir ../training_data/new_data/test \
--val_dir ../training_data/new_data/val \
--combine_dir ../training_data/new_data/combine

1. Run on Jupyter Notebook

%%bash
python3 prepare_data.py