Update app.py

app.py

@@ -4,6 +4,7 @@ import requests
 from io import BytesIO
 import numpy as np
 import json
+import tempfile
 import easyocr
 from transformers import TrOCRProcessor, VisionEncoderDecoderModel
 
@@ -32,7 +33,7 @@ def detect_text_trocr_json(image_file, image_url):
     words_json = []
 
     for bbox, _, conf in results:
-        # Convert coordinates to float
+        # Convert coordinates to float for JSON serialization
         x_coords = [float(point[0]) for point in bbox]
         y_coords = [float(point[1]) for point in bbox]
         x_min, y_min = min(x_coords), min(y_coords)
@@ -52,7 +53,7 @@ def detect_text_trocr_json(image_file, image_url):
             "confidence": float(conf)
         })
 
-    # For simplicity, treat words as paragraphs
+    # Treat words as paragraphs for simplicity
     paragraphs_json = words_json.copy()
 
     output_json = {
@@ -60,8 +61,14 @@ def detect_text_trocr_json(image_file, image_url):
         "paragraphs": paragraphs_json
     }
 
-    json_str = json.dumps(output_json, indent=2)  # now serializable
-    return image, json_str, json_str
+    json_str = json.dumps(output_json, indent=2)
+
+    # Save JSON to a temporary file for Gradio download
+    tmp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".json", mode='w')
+    tmp_file.write(json_str)
+    tmp_file.close()
+
+    return image, json_str, tmp_file.name
 
 iface = gr.Interface(
     fn=detect_text_trocr_json,