diff --git a/changelog.d/20241126_094511_msarniak_fix_creation_time.md b/changelog.d/20241126_094511_msarniak_fix_creation_time.md
new file mode 100644
index 000000000000..7bb1eab6be62
--- /dev/null
+++ b/changelog.d/20241126_094511_msarniak_fix_creation_time.md
@@ -0,0 +1,4 @@
+### Fixed
+
+- Task creation performance for > 50k frames jobs
+  (<https://github.com/cvat-ai/cvat/pull/8741>)
diff --git a/cvat/apps/engine/media_extractors.py b/cvat/apps/engine/media_extractors.py
index 9bee5fbee4cb..6fde4824f4ca 100644
--- a/cvat/apps/engine/media_extractors.py
+++ b/cvat/apps/engine/media_extractors.py
@@ -208,6 +208,7 @@ def __init__(
         dimension: DimensionType = DimensionType.DIM_2D
     ):
         self._source_path = source_path
+        self._source_path_set = set(source_path) if source_path else set()
 
         self._step = step
 
@@ -312,7 +313,7 @@ def __iter__(self):
             yield (self.get_image(i), self.get_path(i), i)
 
     def __contains__(self, media_file):
-        return media_file in self._source_path
+        return media_file in self._source_path_set
 
     def filter(self, callback):
         source_path = list(filter(callback, self._source_path))
diff --git a/cvat/apps/engine/task.py b/cvat/apps/engine/task.py
index 1158bfa700a6..b4d554838817 100644
--- a/cvat/apps/engine/task.py
+++ b/cvat/apps/engine/task.py
@@ -250,7 +250,11 @@ def _count_files(data):
         if not os.path.dirname(v[0]).startswith(v[1])]
 
     # we need to keep the original sequence of files
-    data['server_files'] = [f for f in server_files if f in without_extra_dirs]
+    # Convert without_extra_dirs to a set for O(1) lookups
+    without_extra_dirs_set = set(without_extra_dirs) if without_extra_dirs else set()
+
+    # Filter server_files based on whether they exist in the set
+    data['server_files'] = [f for f in server_files if f in without_extra_dirs_set]
 
     def count_files(file_mapping, counter):
         for rel_path, full_path in file_mapping.items():
@@ -722,9 +726,12 @@ def _update_status(msg: str) -> None:
 
         # We only need to process the files specified in job_file_mapping
         if job_file_mapping is not None:
+            # Convert data['server_files'] to a set for O(1) membership checks
+            server_files_set = set(data['server_files']) if data['server_files'] else set()
+
             filtered_files = []
             for f in itertools.chain.from_iterable(job_file_mapping):
-                if f not in data['server_files']:
+                if f not in server_files_set:
                     raise ValidationError(f"Job mapping file {f} is not specified in input files")
                 filtered_files.append(f)
             data['server_files'] = filtered_files
diff --git a/utils/dataset_manifest/core.py b/utils/dataset_manifest/core.py
index c50c3c470cc6..98bea214cf5b 100644
--- a/utils/dataset_manifest/core.py
+++ b/utils/dataset_manifest/core.py
@@ -682,10 +682,14 @@ def data(self):
     def get_subset(self, subset_names):
         index_list = []
         subset = []
+        # First, create a dictionary mapping image names to their indices
+        name_to_index = {name: index for index, name in enumerate(subset_names)} if subset_names else {}
+
+        # Now, loop through the images and check against the dictionary
         for _, image in self:
             image_name = f"{image.full_name}"
-            if image_name in subset_names:
-                index_list.append(subset_names.index(image_name))
+            if image_name in name_to_index:
+                index_list.append(name_to_index[image_name])
                 properties = {
                     "name": f"{image['name']}",
                     "extension": f"{image['extension']}",