Update app.py

app.py

@@ -7,6 +7,8 @@ from googletrans import Translator
 import cv2
 import numpy as np
 import tempfile
+import base64
+from io import BytesIO 
 
 def trans(text, lang='ta'):
     translator = Translator()
@@ -37,16 +39,27 @@ def object_recognition(image_array, lang):
     return audio_file
 
 def ocr_detection(image_array, lang):
-    # Convert the NumPy array to PIL Image
     image = Image.fromarray(image_array)
 
-    client = Client("https://kneelesh48-tesseract-ocr.hf.space/")
-    result = client.predict(image, "afr", api_name="/tesseract-ocr")
-    print(result)
-    text = "OCR detection result for the captured image."
-    audio_file = trans(result, lang)
+    buffered = BytesIO()
+    image.save(buffered, format="PNG")
+    image_base64 = base64.b64encode(buffered.getvalue()).decode()
+
+    response = requests.post("https://pragnakalp-ocr-image-to-text.hf.space/run/predict", json={
+        "data": [
+            "PaddleOCR",
+            f"data:image/png;base64,{image_base64}",
+        ]
+    }).json()
+
+    data = response.get("data", [])
+
+    text = " ".join(data)  
+    audio_file = trans(text, lang)
+
     return audio_file
 
+
 def operator(image_array, value, lang):
     if value == "1":
         audio_file = object_recognition(image_array, lang)
@@ -59,4 +72,4 @@ def operator(image_array, value, lang):
 
 # Create Gradio interface
 iface = gr.Interface(fn=operator, inputs=["image", "text", "text"], outputs="audio")
-iface.launch(share=True)
+iface.launch(share=True)