MMOCR is compiling-free via moving textdet postprocessing and ops to mmcv 1.3.4 or later.
Add a new OCR downstream task-NER.
Add two new text detection methods: DRRG and FCENet.
Add a new text recognition method: TPS-CRNN.
Add an end-to-end demo.
MMOCR is released.
Support text detection, text recognition and the corresponding downstream tasks such as key information extraction.
For text detection, support both single-step (
TextSnake) and two-step (
For text recognition, support CTC-loss based method
CRNN; Encoder-decoder (with attention) based methods
Robustscanner; Segmentation based method
SegOCR; Transformer based method
For key information extraction, support GCN based method
Provide checkpoints and log files for all of the methods above.