A few days ago, I made an OCR project using CRAFT for detection and CRNN. However, that project's inputs require a point where near the nearest target text aera. This project makes a modification that omit the point input and is able to detect all text in the picture.
the requirements of the project can be found in original OCR project Put the weights of CRAFT and CRNN in the ./weights/ and test images in ./test_data/ Then run
python test.py
The results is saved in ./result/