HyperOCR
A Simple light-weight text detection implementation based on CTPN with mobilenet1.0 backbone.
Startup
python demo_ctpn.py --image textx.jpg --prefix model/rpn1 --epoch 80 --gpu 0 --vis
Demo Image
Training Tips
- Execute script init.sh(init.bat on Windows) to initialize project.
- Download pretrained model from here into
model
folder. - Download dataset from baidu yun. This dataset is already prepared by @eragonruan to fit CTPN.
- Unzip the dataset downloaded to
'VOCdevkit'
folder, and set bothdefault.root_path
anddefault.dataset_path
inrcnn/config.py
to'<somewhere>/VOCdevkit/VOC2007'
. You can also change other hyperparams inrcnn/config.py
. - Run
python train_ctpn.py
to train. Runpython train_ctpn.py --gpus '0' --rpn_lr 0.01 --no_flip 0
to train model on gpu 0 + with learning rate 0.01 and with flip data augmentation.
Tips
- This is an experimental , maybe cannot achieve good effect.
- The code based on https://github.com/chinakook/CTPN.mxnet
TODO
-
new implementation based on PSENET
-
Text region feature extractor for template based ocr.
-
OCR model.