Create app.py

app.py

@@ -0,0 +1,50 @@
+import cv2
+import easyocr
+import gradio as gr
+import base64
+import json
+
+def text_extraction(image):
+    # Convert base64 image to OpenCV format
+    image = base64.b64decode(image.split(",")[1])
+    nparr = np.frombuffer(image, np.uint8)
+    img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
+
+    # Instance text detector
+    reader = easyocr.Reader(['en'], gpu=False)
+
+    # Detect text on image
+    text_ = reader.readtext(img)
+
+    threshold = 0.25
+    # Draw bbox and text
+    for t_, t in enumerate(text_):
+        bbox, text, score = t
+
+        if score > threshold:
+            cv2.rectangle(img, tuple(map(int, bbox[0])), tuple(map(int, bbox[2])), (255, 0, 0), 2)
+
+    # Encode image to base64
+    retval, buffer = cv2.imencode('.jpg', img)
+    img_base64 = base64.b64encode(buffer).decode('utf-8')
+
+    # Create JSON response
+    response_json = {
+        'annotated_image_base64': img_base64,
+        'text_data': text_
+    }
+
+    # Convert the dictionary to a JSON string
+    response_json_str = json.dumps(response_json, default=str)
+
+    return response_json_str
+
+# Define Gradio interface
+iface = gr.Interface(
+    fn=text_extraction,
+    inputs=gr.Image(),
+    outputs=["image", "json"]
+)
+
+# Launch the Gradio interface
+iface.launch()