Update app.py

app.py

@@ -1,9 +1,8 @@
 from PIL import Image
 import torch
 from transformers import NougatProcessor, VisionEncoderDecoderModel
-import gradio as gr
 
-# Load model and processor once at startup
+# Load the model and processor
 processor = NougatProcessor.from_pretrained("MohamedRashad/arabic-small-nougat")
 model = VisionEncoderDecoderModel.from_pretrained("MohamedRashad/arabic-small-nougat")
 
@@ -12,9 +11,9 @@ model.to(device)
 
 context_length = 2048
 
-def predict(image):
-    # Ensure image is in RGB format
-    image = image.convert("RGB")
+def predict(img_path):
+    # Open and ensure RGB format
+    image = Image.open(img_path).convert("RGB")
 
     # Prepare input
     pixel_values = processor(images=image, return_tensors="pt").pixel_values
@@ -33,18 +32,5 @@ def predict(image):
 
     return page_sequence
 
-# Gradio Interface
-title = "Arabic Nougat OCR - Handwritten & Printed Document Recognizer"
-description = "Transcribe Arabic documents using a fine-tuned Nougat model."
-
-interface = gr.Interface(
-    fn=predict,
-    inputs=gr.Image(type="pil", label="Upload an Arabic Document"),
-    outputs=gr.Textbox(label="Transcription", lines=15),
-    title=title,
-    description=description,
-    examples=[["example_1.jpg"], ["example_2.jpg"]]
-)
-
-if __name__ == "__main__":
-    interface.launch()
+# Test the OCR
+print(predict("1.png"))