A Unet based deeplearning model to line/box/spurious artifacts from text images. Unsupervised training.

patchwise semantic segmentation

This tool can detect textline of given image and write the output in a page xml data.

Document Layout Analysis repos for development with PdfPig.

FreeAnchor: Learning to Match Anchors for Visual Object Detection (NeurIPS 2019)

An Implementation of the FOTS: Fast Oriented Text Spotting with a Unified Network

TensorFlow implementation of a segmentation system for document images.

This is a tool for pixel wise segmentation. This is developed in order to do use cases like page extraction, textline , word and structure recognition of library documents.

Adversarial Generation of Handwritten Text Images

very fast CTPN

Combine Textboxes++ and RetinaNet which used to replace the original 'neck' network SSD for scene text localization.

Code and data for the paper at http://arxiv.org/abs/2004.07317

Distorted Document Images dataset (DDI-100).

Code for Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells, CVPR '19

FCDenseNet implementation for Table Localization

PKU Team Zero's code for participation in ICDAR2019 ArT Recognition track (Champion)

This repository provides train＆test code, dataset, det.&rec. annotation, evaluation script, annotation tool, and ranking table.

DANet: Divergent Activation for Weakly Supervised Object Localization，in ICCV 2019

In-depth tutorials for implementing deep learning models on your own with PyTorch.