diff --git a/spotiflow/cli/predict.py b/spotiflow/cli/predict.py
index 0a518e6..0f99e7e 100644
--- a/spotiflow/cli/predict.py
+++ b/spotiflow/cli/predict.py
@@ -70,9 +70,19 @@ def get_args():
                             type=str, required=False, default="auto", choices=["auto", "cpu", "cuda", "mps"],
                             help="Device to run model on. Defaults to 'auto'.")
 
+    utils = parser.add_argument_group(title="Utility arguments",
+                                      description="Diverse utility arguments, e.g. I/O related.")
+    utils.add_argument("--exclude-hidden-files", action="store_true", required=False, default=False, help="Exclude hidden files in the input directory. Defaults to False.")
+
     args = parser.parse_args()
     return args
 
+def _imread_wrapped(fname):
+    try:
+        return imread(fname)
+    except Exception as e:
+        log.error(f"Could not read image {fname}. Execution will halt.")
+        raise e
 
 def main():
     # Get arguments from command line
@@ -108,6 +118,8 @@ def main():
         image_files = sorted(
             tuple(chain(*tuple(args.data_path.glob(f"*.{ext}") for ext in ALLOWED_EXTENSIONS)))
         )
+        if args.exclude_hidden_files:
+            image_files = tuple(f for f in image_files if not f.name.startswith("."))
         if len(image_files) == 0:
             raise ValueError(f"No valid image files found in directory {args.data_path}. Allowed extensions are: {ALLOWED_EXTENSIONS}")
         if out_dir is None:
@@ -119,7 +131,7 @@ def main():
     out_dir.mkdir(exist_ok=True, parents=True)
 
     # Predict spots in images and write to CSV
-    images = [imread(img) for img in image_files]
+    images = [_imread_wrapped(img) for img in image_files]
     for img, fname in tqdm(zip(images, image_files), desc="Predicting", total=len(images)):
         spots, _ = model.predict(img,
                                  prob_thresh=args.probability_threshold,