Paper Implement on this https://jodac.org/u-net-convolutional-neural-network-for-binarization-of-historical-kannada-handwritten-palm-leaf-manuscripts/

LS-HDIB Dataset

LS-HDIB: A Large Scale Handwritten Document Image Binarization Dataset

Link to download initial dataset is here.

Setup and Install

git clone https://github.com/kaustubh-sadekar/LS-HDIB.git
cd LS-HDIB/
pip install -r requirements.txt

Run segmentation for your own image

python run.py cpu input_2.jpg unet_best_weights.pth

The output file will be saved as <INPUT_FILE_NAME>_output.jpg. For this specific example it will be input_2_output.jpg

For better understanding of the input arguments type python run.py -h

Coming Soon

Easy to use google colab notebook

NOTE:

We used segmentation_models_pytorch library for all the segmentation models. It has implementations for several segmentation models.