A project to recognize text in Vietnamese ID card using Centernet-Custom and Seq2Seq-Attention models
Edit requirements_gpu.txt or requirements_cpu.txt depend on CPU or GPU, then run:
sh setup.sh
Model | mAP(IOU=0.5-0.95) | Full Sequence Acc (%) |
---|---|---|
CMND Detect |
x | x |
Line Detect |
0.75 | x |
OCR |
x | 81% |
Run demo from file OCR_deploy.ipynb