added GPU config

IndicPhotoOCR/ocr.py

@@ -121,7 +121,7 @@ class OCR:
         """Recognize text in a cropped image area using the identified script."""
         if self.verbose:
             print("Recognizing text in detected area...")
-        recognized_text = self.recogniser.recognise(script_lang, cropped_image_path, script_lang, self.verbose)
+        recognized_text = self.recogniser.recognise(script_lang, cropped_image_path, script_lang, self.verbose, self.device)
         # print(recognized_text)
         return recognized_text
 

IndicPhotoOCR/recognition/parseq_recogniser.py

@@ -188,7 +188,7 @@ class PARseqrecogniser:
         return text
 
 
-    def recognise(self, checkpoint: str, image_path: str, language: str, verbose: bool) -> str:
+    def recognise(self, checkpoint: str, image_path: str, language: str, verbose: bool, device: str) -> str:
         """
         Loads the desired model and returns the recognized word from the specified image.
 
@@ -200,7 +200,7 @@ class PARseqrecogniser:
         Returns:
             str: The recognized text from the image.
         """
-        device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
+        # device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
         if language != "english":
             model_path = self.ensure_model(checkpoint)

app.py

@@ -6,7 +6,7 @@ from IndicPhotoOCR.theme import Seafoam
 from IndicPhotoOCR.utils.helper import detect_para
 
 # Initialize the OCR object for text detection and recognition
-ocr = OCR(device="cpu", verbose=False)
+ocr = OCR(verbose=False)
 
 def process_image(image):
     """
@@ -129,7 +129,7 @@ demo = gr.Interface(
 # # Server setup and launch configuration
 # if __name__ == "__main__":
 #     server = "0.0.0.0"  # IP address for server
-#     port = 7865  # Port to run the server on
+#     port = 7866  # Port to run the server on
 #     demo.launch(server_name=server, server_port=port, share=True)
 
 demo.launch()