a micro OCR network.
This model can handle complex tasks without lstm, and its accuracy and speed are better than resnet and crnn models.
MicroOCR
├── README.md # descriptions about MicroNet
├── average_meter.py # average meter
├── collatefn.py # batch data processing
├── label_converter.py # label converter
├── dataset.py # data preprocessing for training and evaluation
├── demo.py # inference
├── gen_image.py # generate image for train and eval
├── img_aug.py # img augmentation
├── infer_tool.py # inference tool
├── logger.py # logger
├── loss.py # ctcloss definition
├── model.py # MicroMLPNet
├── train.py # train the model
python gen_image.py
python train.py
python demo.py