Update app.py
Browse files
app.py
CHANGED
@@ -4,6 +4,7 @@ from PIL import Image, ImageDraw, ImageFont
|
|
4 |
import json
|
5 |
from paddleocr import PaddleOCR
|
6 |
import gradio as gr
|
|
|
7 |
|
8 |
# Initialize PaddleOCR
|
9 |
ocr = PaddleOCR(use_angle_cls=True, lang='en')
|
@@ -27,10 +28,10 @@ def draw_boxes_on_image(image, data):
|
|
27 |
bounding_box, (text, confidence) = item
|
28 |
|
29 |
# Ensure bounding_box is a list of lists
|
30 |
-
if isinstance(bounding_box[0], list):
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
|
35 |
# Draw the bounding box
|
36 |
draw.line([tuple(box[0]), tuple(box[1])], fill="green", width=2)
|
@@ -44,26 +45,29 @@ def draw_boxes_on_image(image, data):
|
|
44 |
|
45 |
return pil_image
|
46 |
|
47 |
-
# Function to
|
48 |
-
def
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
|
|
|
|
|
|
67 |
|
68 |
# Function to identify 'field', 'value' pairs
|
69 |
def identify_field_value_pairs(ocr_results, fields):
|
@@ -86,16 +90,15 @@ def process_image(image):
|
|
86 |
processed_image = draw_boxes_on_image(image, ocr_results[0])
|
87 |
|
88 |
# Save OCR results to JSON
|
89 |
-
results_json = save_results_to_json(ocr_results[0])
|
90 |
json_path = "ocr_results.json"
|
91 |
-
|
92 |
-
json.dump(results_json, json_file, indent=4)
|
93 |
|
94 |
# Identify field-value pairs
|
95 |
fields = ["Scheme Name", "Folio Number", "Number of Units", "PAN", "Signature", "Tax Status",
|
96 |
"Mobile Number", "Email", "Address", "Bank Account Details"]
|
97 |
field_value_pairs = identify_field_value_pairs(ocr_results[0], fields)
|
98 |
field_value_json_path = "field_value_pairs.json"
|
|
|
99 |
with open(field_value_json_path, 'w') as json_file:
|
100 |
json.dump(field_value_pairs, json_file, indent=4)
|
101 |
|
@@ -107,8 +110,8 @@ interface = gr.Interface(
|
|
107 |
inputs="image",
|
108 |
outputs=[
|
109 |
"image",
|
110 |
-
gr.File(label="OCR Results JSON"),
|
111 |
-
gr.File(label="Field-Value Pairs JSON")
|
112 |
],
|
113 |
title="OCR Web Application",
|
114 |
description="Upload an image and get OCR results with bounding boxes and two JSON outputs."
|
|
|
4 |
import json
|
5 |
from paddleocr import PaddleOCR
|
6 |
import gradio as gr
|
7 |
+
import os
|
8 |
|
9 |
# Initialize PaddleOCR
|
10 |
ocr = PaddleOCR(use_angle_cls=True, lang='en')
|
|
|
28 |
bounding_box, (text, confidence) = item
|
29 |
|
30 |
# Ensure bounding_box is a list of lists
|
31 |
+
if not isinstance(bounding_box[0], list):
|
32 |
+
bounding_box = [bounding_box]
|
33 |
+
|
34 |
+
box = np.array(bounding_box).astype(int)
|
35 |
|
36 |
# Draw the bounding box
|
37 |
draw.line([tuple(box[0]), tuple(box[1])], fill="green", width=2)
|
|
|
45 |
|
46 |
return pil_image
|
47 |
|
48 |
+
# Function to convert OCR results to JSON
|
49 |
+
def convert_to_json(results, output_file):
|
50 |
+
"""
|
51 |
+
Converts the given results into a JSON file.
|
52 |
+
|
53 |
+
Args:
|
54 |
+
results: The list of results containing bounding box coordinates, text, and confidence.
|
55 |
+
output_file: The name of the output JSON file.
|
56 |
+
"""
|
57 |
+
json_data = []
|
58 |
+
for result in results:
|
59 |
+
bounding_box = result[0]
|
60 |
+
text = result[1][0]
|
61 |
+
confidence = result[1][1]
|
62 |
+
|
63 |
+
json_data.append({
|
64 |
+
"bounding_box": [list(map(float, coord)) for coord in bounding_box],
|
65 |
+
"text": text,
|
66 |
+
"confidence": confidence
|
67 |
+
})
|
68 |
+
|
69 |
+
with open(output_file, "w") as f:
|
70 |
+
json.dump(json_data, f, indent=4)
|
71 |
|
72 |
# Function to identify 'field', 'value' pairs
|
73 |
def identify_field_value_pairs(ocr_results, fields):
|
|
|
90 |
processed_image = draw_boxes_on_image(image, ocr_results[0])
|
91 |
|
92 |
# Save OCR results to JSON
|
|
|
93 |
json_path = "ocr_results.json"
|
94 |
+
convert_to_json(ocr_results[0], json_path)
|
|
|
95 |
|
96 |
# Identify field-value pairs
|
97 |
fields = ["Scheme Name", "Folio Number", "Number of Units", "PAN", "Signature", "Tax Status",
|
98 |
"Mobile Number", "Email", "Address", "Bank Account Details"]
|
99 |
field_value_pairs = identify_field_value_pairs(ocr_results[0], fields)
|
100 |
field_value_json_path = "field_value_pairs.json"
|
101 |
+
|
102 |
with open(field_value_json_path, 'w') as json_file:
|
103 |
json.dump(field_value_pairs, json_file, indent=4)
|
104 |
|
|
|
110 |
inputs="image",
|
111 |
outputs=[
|
112 |
"image",
|
113 |
+
gr.File(label="Download OCR Results JSON"),
|
114 |
+
gr.File(label="Download Field-Value Pairs JSON")
|
115 |
],
|
116 |
title="OCR Web Application",
|
117 |
description="Upload an image and get OCR results with bounding boxes and two JSON outputs."
|