Update app.py

app.py

@@ -5,54 +5,87 @@ import easyocr
 from PIL import Image
 import numpy as np
 from transformers import MarianMTModel, MarianTokenizer
+import io
 
-# Cloudinary configuration
+# ---------------------
+# Cloudinary Config (replace with your actual values)
+# ---------------------
 cloudinary.config(
-    cloud_name="your_cloud_name",
-    api_key="your_api_key",
-    api_secret="your_api_secret"
+    cloud_name="deoux7285",        # <-- 🔁 Replace this
+    api_key="494111995449977",              # <-- 🔁 Replace this
+
 )
 
-# Initialize OCR reader
+# ---------------------
+# Load OCR & Translation Models
+# ---------------------
 reader = easyocr.Reader(['en'])
 
-# Initialize translation model
-model_name = 'Helsinki-NLP/opus-mt-en-hin'
+# Using English to Hindi as example
+model_name = "Helsinki-NLP/opus-mt-en-hi"
 tokenizer = MarianTokenizer.from_pretrained(model_name)
 model = MarianMTModel.from_pretrained(model_name)
 
-def translate_text(text, target_lang='hin'):
-    translated = model.generate(**tokenizer.prepare_seq2seq_batch(text, return_tensors="pt"))
+# ---------------------
+# Function to Translate Text
+# ---------------------
+def translate_text(text):
+    batch = tokenizer.prepare_seq2seq_batch([text], return_tensors="pt")
+    translated = model.generate(**batch)
     return tokenizer.decode(translated[0], skip_special_tokens=True)
 
-def process_image(image, target_lang='hin'):
-    # OCR to extract text
-    result = reader.readtext(np.array(image))
-    extracted_text = " ".join([item[1] for item in result])
+# ---------------------
+# Main Function
+# ---------------------
+def process_image(image, font_size):
+    # Step 1: OCR
+    np_image = np.array(image)
+    results = reader.readtext(np_image)
+    if not results:
+        return "No text found", "No text extracted."
+
+    extracted_text = " ".join([res[1] for res in results])
 
-    # Translate extracted text
-    translated_text = translate_text(extracted_text, target_lang)
+    # Step 2: Translate
+    translated_text = translate_text(extracted_text)
 
-    # Upload image to Cloudinary
+    # Step 3: Upload to Cloudinary
     buffered = io.BytesIO()
     image.save(buffered, format="JPEG")
     buffered.seek(0)
     upload_result = cloudinary.uploader.upload(buffered)
-    cloud_url = upload_result['secure_url']
+    public_id = upload_result["public_id"]
 
-    # Apply text overlay using Cloudinary
-    overlay_url = f"{cloud_url}?overlay_text={translated_text}&font_size=30&font_color=white&gravity=center"
+    # Step 4: Create Cloudinary overlay URL
+    # NOTE: Overlay text needs to be URL-safe (spaces → %20)
+    safe_text = translated_text.replace(" ", "%20")
+    overlay_url = (
+        f"https://res.cloudinary.com/{cloudinary.config().cloud_name}/image/upload/"
+        f"l_text:arial_{font_size}:{safe_text},g_center,c_scale/"
+        f"{public_id}.jpg"
+    )
 
     return overlay_url, translated_text
 
-# Gradio interface
+# ---------------------
+# Gradio Interface
+# ---------------------
 iface = gr.Interface(
     fn=process_image,
-    inputs=[gr.Image(type="pil", label="Upload Image"), gr.Textbox(label="Target Language (e.g., hin)"), gr.Slider(minimum=10, maximum=100, label="Font Size")],
-    outputs=[gr.Image(label="Image with Overlay"), gr.Textbox(label="Translated Text")],
-    title="Image Translator with Overlay",
-    description="Upload an image, extract text using OCR, translate it, and overlay the translated text on the image."
+    inputs=[
+        gr.Image(type="pil", label="Upload Image"),
+        gr.Slider(20, 80, value=30, label="Overlay Font Size")
+    ],
+    outputs=[
+        gr.Image(type="filepath", label="Image with Translated Text"),
+        gr.Textbox(label="Translated Text")
+    ],
+    title="Image Translator with Text Overlay (Cloudinary)",
+    description="Upload an image, extract English text, translate it to Hindi, and overlay it on the image using Cloudinary."
 )
 
+# ---------------------
+# Launch App
+# ---------------------
 if __name__ == "__main__":
     iface.launch(share=True)