Extracting Text from US Driving Licence Using CRAFT and Deep Text Recognition
The Code is in initial stage.
- lmdb==0.98
- natsort==7.0.1
- nltk==3.5
- numpy==1.18.3
- opencv-contrib-python==4.2.0.34
- Pillow==7.1.2
- pip==20.1
- scikit-image==0.16.2
- scipy==1.4.1
- setuptools==39.0.1
- six==1.14.0
- torch==1.5.0+cpu
- torchvision==0.6.0+cpu
- wheel==0.34.2
pip install -r requirements.txt
Download Pre-trained Model/Weight
- craft_mlt_25k.pth,
- TPS-ResNet-BiLSTM-Attn.pth
Run the below code:
python extract.py -i data/AL.jpeg -o result
where
"extract.py" -> filename
"data/AL.jpeg" -> data is input directory and AL.jpeg is image file
"result" -> output directory where extracted text will be saved
This implementation has been based on these repository clovaai/CRAFT-pytorch, clovaai/deep-text-recognition-benchmark.
- WebDemo : https://demo.ocr.clova.ai/
Combination of Clova AI detection and recognition, additional/advanced features used for KOR/JPN.