Update app.py

app.py

@@ -1,38 +1,59 @@
 import gradio as gr
 import easyocr
 from deep_translator import GoogleTranslator
+from PIL import Image, ImageDraw, ImageFont
+import numpy as np
 
-reader = easyocr.Reader(['en', 'hi'])  # Hindi aur English OCR ke liye
-
-def extract_and_translate(image, target_lang):
-    result = reader.readtext(image, detail=0)
-    text = ' '.join(result)
-    try:
-        translated_text = GoogleTranslator(source='auto', target=target_lang).translate(text)
-    except Exception as e:
-        translated_text = f"Translation Error: {str(e)}"
-    return translated_text
-
-lang_choices = {
-    "Hindi": "hi",
-    "English": "en",
-    "Punjabi": "pa",
-    "Gujarati": "gu",
-    "Bengali": "bn",
-    "Tamil": "ta",
-    "Telugu": "te",
-    "Marathi": "mr"
-}
+# Initialize reader with allowed lang_list only
+reader = easyocr.Reader(['en', 'hi', 'mr', 'ne'])
 
+# Font path — works on HuggingFace
+FONT_PATH = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
+
+def process_image(image, target_lang):
+    img_array = np.array(image)
+
+    # Run OCR safely
+    results = reader.readtext(img_array)
+
+    if not results:
+        return image, "No text detected"
+
+    final_translations = []
+    img_pil = image.convert("RGB")
+    draw = ImageDraw.Draw(img_pil)
+    font = ImageFont.truetype(FONT_PATH, size=20)
+
+    for (bbox, text, prob) in results:
+        if text.strip():
+            try:
+                translated = GoogleTranslator(source='auto', target=target_lang).translate(text)
+            except:
+                translated = "[Translation Error]"
+
+            final_translations.append(translated)
+
+            # Draw translation at top-left of bounding box
+            top_left = bbox[0]
+            draw.text((top_left[0], top_left[1] - 20), translated, fill=(255, 0, 0), font=font)
+
+    full_text = "\n".join(final_translations)
+    return img_pil, full_text
+
+# Gradio UI
 iface = gr.Interface(
-    fn=extract_and_translate,
+    fn=process_image,
     inputs=[
-        gr.Image(type="filepath", label="Upload Image"),
-        gr.Dropdown(choices=list(lang_choices.values()), label="Translate To")
+        gr.Image(type="pil", label="Upload Image"),
+        gr.Textbox(label="Translate To (hi / en / mr / ne)")
+    ],
+    outputs=[
+        gr.Image(type="pil", label="Image with Translated Overlay"),
+        gr.Textbox(label="Translated Text Output")
     ],
-    outputs="text",
-    title="Image Text Translator",
-    description="Upload an image with printed text. Extracts and translates text into selected language."
+    title="Image Translator with Overlay",
+    description="Extract & translate text from an image (OCR + translation overlay)"
 )
 
-iface.launch()
+if __name__ == "__main__":
+    iface.launch()