shreyasvaidya
commited on
Upload folder using huggingface_hub
Browse files
app.py
CHANGED
@@ -11,6 +11,12 @@ from transformers import (
|
|
11 |
)
|
12 |
import numpy as np
|
13 |
import torch
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
from IndicTransToolkit import IndicProcessor
|
16 |
|
@@ -107,6 +113,7 @@ def process_image(image):
|
|
107 |
# # Recognize text in the cropped area
|
108 |
# recognized_text = ocr.recognise(cropped_path, script_lang)
|
109 |
# recognized_texts.append(recognized_text)
|
|
|
110 |
for id, bbox in enumerate(detections):
|
111 |
# Identify the script and crop the image to this region
|
112 |
script_lang, cropped_path = ocr.crop_and_identify_script(pil_image, bbox)
|
@@ -120,13 +127,14 @@ def process_image(image):
|
|
120 |
if script_lang:
|
121 |
recognized_text = ocr.recognise(cropped_path, script_lang)
|
122 |
recognized_texts[f"img_{id}"] = {"txt": recognized_text, "bbox": [x1, y1, x2, y2]}
|
|
|
123 |
|
124 |
# Combine recognized texts into a single string for display
|
125 |
# recognized_texts_combined = " ".join(recognized_texts)
|
126 |
string = detect_para(recognized_texts)
|
127 |
|
128 |
recognized_texts_combined = '\n'.join([' '.join(line) for line in string])
|
129 |
-
recognized_texts_combined = translate(recognized_texts_combined,
|
130 |
|
131 |
return output_image, recognized_texts_combined
|
132 |
|
|
|
11 |
)
|
12 |
import numpy as np
|
13 |
import torch
|
14 |
+
from collections import Counter
|
15 |
+
|
16 |
+
def Most_Common(lst):
|
17 |
+
data = Counter(lst)
|
18 |
+
return data.most_common(1)[0][0]
|
19 |
+
|
20 |
|
21 |
from IndicTransToolkit import IndicProcessor
|
22 |
|
|
|
113 |
# # Recognize text in the cropped area
|
114 |
# recognized_text = ocr.recognise(cropped_path, script_lang)
|
115 |
# recognized_texts.append(recognized_text)
|
116 |
+
langs = []
|
117 |
for id, bbox in enumerate(detections):
|
118 |
# Identify the script and crop the image to this region
|
119 |
script_lang, cropped_path = ocr.crop_and_identify_script(pil_image, bbox)
|
|
|
127 |
if script_lang:
|
128 |
recognized_text = ocr.recognise(cropped_path, script_lang)
|
129 |
recognized_texts[f"img_{id}"] = {"txt": recognized_text, "bbox": [x1, y1, x2, y2]}
|
130 |
+
langs.append(script_lang)
|
131 |
|
132 |
# Combine recognized texts into a single string for display
|
133 |
# recognized_texts_combined = " ".join(recognized_texts)
|
134 |
string = detect_para(recognized_texts)
|
135 |
|
136 |
recognized_texts_combined = '\n'.join([' '.join(line) for line in string])
|
137 |
+
recognized_texts_combined = translate(recognized_texts_combined,Most_Common(langs))
|
138 |
|
139 |
return output_image, recognized_texts_combined
|
140 |
|