A collection of original, innovative ideas and algorithms towards Advanced Literate Machinery. This project is maintained by the OCR Team in the Language Technology Lab, Tongyi Lab, Alibaba Group.

TextBoxes++: A Single-Shot Oriented Scene Text Detector

Scene text detection and recognition based on Extremal Region(ER)

End-to-end pipeline for real-time scene text detection and recognition.

TensorFlow Implementation of FOTS, Fast Oriented Text Spotting with a Unified Network.

This is a c++ project deploying a deep scene text reading pipeline with tensorflow. It reads text from natural scene images. It uses frozen tensorflow graphs. The detector detect scene text locations. The recognizer reads word from each detected bounding box.

Implement 'Image deformation using moving least squares' with pybind11 and python3 and opencv2

A high speed character recognition engine

Three different ways to Detect, Recognize or Spot Scene Text. Done for GeoAI Internship at GeoWorks, Singapore.