|
import os |
|
from doctr.io import DocumentFile |
|
from doctr.models import ocr_predictor, from_hub |
|
import gradio as gr |
|
|
|
os.environ['USE_TORCH'] = '1' |
|
reco_model_zgh = from_hub('ayymen/crnn_mobilenet_v3_large_zgh') |
|
predictor_zgh = ocr_predictor(reco_arch=reco_model_zgh, pretrained=True) |
|
|
|
reco_model = from_hub('ayymen/crnn_mobilenet_v3_large_tifinagh') |
|
predictor = ocr_predictor(reco_arch=reco_model, pretrained=True) |
|
|
|
title = "Tifinagh OCR" |
|
description = """Upload an image to get the OCR results! |
|
""" |
|
|
|
def ocr(img, script): |
|
img.save("out.jpg") |
|
doc = DocumentFile.from_images("out.jpg") |
|
output = predictor_zgh(doc) if script == "Tifinagh-IRCAM" else predictor(doc) |
|
res = "" |
|
for obj in output.pages: |
|
for obj1 in obj.blocks: |
|
for obj2 in obj1.lines: |
|
for obj3 in obj2.words: |
|
res = res + " " + obj3.value |
|
res = res + "\n" |
|
res = res + "\n" |
|
_output_name = "RESULT_OCR.txt" |
|
open(_output_name, 'w', encoding="utf-8").close() |
|
with open(_output_name, "w", encoding="utf-8", errors="ignore") as f: |
|
f.write(res) |
|
print("Writing into file") |
|
return res, _output_name |
|
|
|
demo = gr.Interface(fn=ocr, |
|
inputs=[ |
|
gr.Image(type="pil"), |
|
gr.Dropdown(choices=['Tifinagh-IRCAM', 'Tifinagh'], label="Script", value="Tifinagh-IRCAM") |
|
], |
|
outputs=[ |
|
gr.Textbox(lines=20, label="Full Text"), |
|
gr.File(label="Download OCR Results") |
|
], |
|
title=title, |
|
description=description, |
|
examples=[ |
|
["Examples/3.jpg", "Tifinagh-IRCAM"], |
|
["Examples/2.jpg", "Tifinagh-IRCAM"], |
|
["Examples/1.jpg", "Tifinagh-IRCAM"] |
|
] |
|
) |
|
|
|
demo.launch(debug=True) |
|
|
