Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,26 +1,59 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
import easyocr
|
| 3 |
import numpy as np
|
|
|
|
| 4 |
|
| 5 |
-
# Initialize EasyOCR reader once with English
|
| 6 |
reader = easyocr.Reader(['en'], gpu=False)
|
| 7 |
|
| 8 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
if img is None:
|
| 10 |
-
return "No image uploaded"
|
|
|
|
|
|
|
| 11 |
img_array = np.array(img)
|
|
|
|
| 12 |
try:
|
| 13 |
-
#
|
| 14 |
results = reader.readtext(img_array, detail=0, paragraph=True)
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
except Exception as e:
|
| 17 |
-
|
|
|
|
| 18 |
|
| 19 |
with gr.Blocks() as demo:
|
| 20 |
-
gr.Markdown("# EasyOCR
|
| 21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
output_text = gr.Textbox(label="OCR Result", lines=10)
|
| 23 |
-
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
demo.launch()
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import easyocr
|
| 3 |
import numpy as np
|
| 4 |
+
from difflib import SequenceMatcher
|
| 5 |
|
| 6 |
+
# Initialize EasyOCR reader once with English
|
| 7 |
reader = easyocr.Reader(['en'], gpu=False)
|
| 8 |
|
| 9 |
+
def calculate_similarity(text1, text2):
|
| 10 |
+
# Calculate similarity ratio for accuracy and pipeline score
|
| 11 |
+
return SequenceMatcher(None, text1.lower().strip(), text2.lower().strip()).ratio()
|
| 12 |
+
|
| 13 |
+
def ocr_easy_with_accuracy(img, correct_text):
|
| 14 |
if img is None:
|
| 15 |
+
return "No image uploaded", "", ""
|
| 16 |
+
|
| 17 |
+
# Convert uploaded image (PIL) to numpy array
|
| 18 |
img_array = np.array(img)
|
| 19 |
+
|
| 20 |
try:
|
| 21 |
+
# Run EasyOCR with paragraph grouping and no detailed bounding boxes
|
| 22 |
results = reader.readtext(img_array, detail=0, paragraph=True)
|
| 23 |
+
detected_text = "\n".join(results)
|
| 24 |
+
|
| 25 |
+
if correct_text.strip():
|
| 26 |
+
accuracy = calculate_similarity(detected_text, correct_text)
|
| 27 |
+
pipeline_score = accuracy # Using same metric for demo
|
| 28 |
+
|
| 29 |
+
accuracy_str = f"{accuracy:.2%}"
|
| 30 |
+
pipeline_score_str = f"{pipeline_score:.2%}"
|
| 31 |
+
else:
|
| 32 |
+
accuracy_str = "No ground truth provided"
|
| 33 |
+
pipeline_score_str = "No ground truth provided"
|
| 34 |
+
|
| 35 |
+
return detected_text, accuracy_str, pipeline_score_str
|
| 36 |
+
|
| 37 |
except Exception as e:
|
| 38 |
+
error_msg = f"EasyOCR Error: {str(e)}"
|
| 39 |
+
return error_msg, "", ""
|
| 40 |
|
| 41 |
with gr.Blocks() as demo:
|
| 42 |
+
gr.Markdown("# EasyOCR with Accuracy and Pipeline Score")
|
| 43 |
+
|
| 44 |
+
with gr.Row():
|
| 45 |
+
img_input = gr.Image(type="pil", label="Upload Image")
|
| 46 |
+
correct_text_input = gr.Textbox(label="Enter Correct Text for Accuracy Check", lines=4)
|
| 47 |
+
|
| 48 |
output_text = gr.Textbox(label="OCR Result", lines=10)
|
| 49 |
+
accuracy_output = gr.Textbox(label="Accuracy", interactive=False)
|
| 50 |
+
pipeline_output = gr.Textbox(label="Pipeline Integration Score", interactive=False)
|
| 51 |
+
|
| 52 |
+
run_button = gr.Button("Run OCR and Check Accuracy")
|
| 53 |
+
run_button.click(
|
| 54 |
+
ocr_easy_with_accuracy,
|
| 55 |
+
inputs=[img_input, correct_text_input],
|
| 56 |
+
outputs=[output_text, accuracy_output, pipeline_output]
|
| 57 |
+
)
|
| 58 |
|
| 59 |
demo.launch()
|