This project contains tools for training PyTorch models for use with the Ocrs OCR engine.
The ocrs engine splits text detection and recognition into three phases, each of which corresponds to a different model in this repository:
All models can be exported to ONNX for downstream use.
The models are trained exclusively on datasets which are a) open and b) have non-restrictive licenses. This currently includes:
Pre-trained models are available from Hugging Face as PyTorch checkpoints, ONNX and RTen models.
See the Training guide for a walk-through of the process to train models from scratch or fine-tune existing models.