Add config_file arg

preprocessing/main.py

@@ -13,9 +13,6 @@ from multiprocessing import cpu_count
 # Configure logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
 
-cambrianConfig = CambrianConfig.from_json_file("config.json")
-processor = CambrianEncoders(cambrianConfig)
-
 def get_optimal_workers() -> int:
     """Determine the optimal number of workers based on available CPU cores."""
     try:
@@ -23,7 +20,7 @@ def get_optimal_workers() -> int:
     except (NotImplementedError, ValueError):
         return 1  # Fallback to a single worker in case of an error
 
-def extract_features(file_path: str, file_name: str) -> Dict[str, torch.Tensor]:
+def extract_features(processor: CambrianEncoders, file_path: str, file_name: str) -> Dict[str, torch.Tensor]:
     try:
         video, image_sizes = process_video_frames(file_path)
         image_aux_features_list = processor.prepare_mm_features(images=video, image_sizes=image_sizes)
@@ -46,12 +43,21 @@ if __name__ == "__main__":
     )
     parser.add_argument(
         '--output_file',
-        type=str,
-        default='entube_tensors.safetensors',
-        help='Safetensor file to store embeddings of EnTube dataset by vision encoders'
+        type = str,
+        default = 'entube_tensors.safetensors',
+        help = 'Safetensor file to store embeddings of EnTube dataset by vision encoders'
+    )
+    parser.add_argument(
+        '--config_file',
+        type = str,
+        default = 'config.json',
+        help = 'Path to configuration file of encoders parameters'
     )
     args = parser.parse_args()
 
+    cambrianConfig = CambrianConfig.from_json_file(args.config_file)
+    processor = CambrianEncoders(cambrianConfig)
+
     folder_paths: List[str] = args.folders
     data_tensor = defaultdict(torch.Tensor)  # Use defaultdict for thread safety
     # Determine optimal workers
@@ -65,7 +71,7 @@ if __name__ == "__main__":
             file_names = os.listdir(folder_path)
             for file_name in file_names:
                 file_path = os.path.join(folder_path, file_name)
-                futures.append(executor.submit(extract_features, file_path, file_name))
+                futures.append(executor.submit(extract_features, processor, file_path, file_name))
 
         # Collect results as tasks complete
         for future in as_completed(futures):