Text Detection with EAST and other methods #13847
Replies: 1 comment
-
Configs file: Global:
Architecture: Loss: Optimizer: PostProcess: Metric: Train: Eval: |
Beta Was this translation helpful? Give feedback.
-
Configs file: Global:
Architecture: Loss: Optimizer: PostProcess: Metric: Train: Eval: |
Beta Was this translation helpful? Give feedback.
-
Dear Paddle Community,
I have performed fine-tuning with en_PP-OCRv3_det with ch_PP-OCRv3_det_student.yml as config file. It has been attached below for your reference.
Fine-tune configs file: https://github.com/PaddlePaddle/PaddleOCR/tree/main/configs/det/ch_PP-OCRv3
It is known that PaddleOCR has 4 text detection algorithms namely EAST, DB, SAST, PSENet.
The default config file uses DB for text detection. I would like to train with EAST now to see if it performs better hence I looked into configs file: https://github.com/PaddlePaddle/PaddleOCR/tree/main/configs/det
Issue:
Link for full training: https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_en/detection_en.md
Query: I humbly request for guidance on how to train text detection with EAST algorithm. Also, which weights to use for it?
Still I am unable to run it like this: ocr = PaddleOCR(use_gpu=True, det_algorithm='EAST',
det_model_dir='C:/Users/Adhi/paddle_fine_tuning/text_detection/05_det_EAST_inference',
use_angle_cls=True, lang='en')
Apparently only DB can be used as an argument.
I checked for alternative: https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_en/inference_en.md
But I do not want to run like this: python3 tools/infer/predict_det.py --image_dir="./doc/imgs_en/img_10.jpg" --det_model_dir="./inference/det_east/" --det_algorithm="EAST"
rather in an integrated way like this: ocr = PaddleOCR(use_gpu=True, det_algorithm='EAST',
det_model_dir='C:/Users/Adhi/paddle_fine_tuning/text_detection/05_det_EAST_inference',
use_angle_cls=True, lang='en'), how to accomplish this?
Thanks in advance!
Best Regards,
Adhitya
Beta Was this translation helpful? Give feedback.
All reactions