MMOCR is an open-source toolbox part of the OpenMMLab project that provides tools for Optical Character Recognition tasks. It includes implementations of various deep learning architectures such as CRNN, DBNet, PSENet, and ABCNet for text detection, recognition, and end-to-end understanding. The toolbox is built on PyTorch and MMCV, supporting model training, evaluation, and inference workflows.