[save-images] Add new config option for setting threading mode

Breakthrough · Nov 24, 2024 · a9f012e · a9f012e
1 parent 55dec7f
commit a9f012e
Show file tree

Hide file tree

Showing 8 changed files with 124 additions and 84 deletions.
diff --git a/scenedetect.cfg b/scenedetect.cfg
@@ -220,22 +220,25 @@
 # Image quality (jpeg/webp). Default is 95 for jpeg, 100 for webp
 #quality = 95
 
-# Compression amount for png images (0 to 9). Does not affect quality.
+# Compression amount for png images (0 to 9). Only affects size, not quality.
 #compression = 3
 
-# Number of frames to skip at beginning/end of scene.
+# Number of frames to ignore around each scene cut when selecting frames.
 #frame-margin = 1
 
-# Factor to resize images by (0.5 = half, 1.0 = same, 2.0 = double).
+# Resize by scale factor (0.5 = half, 1.0 = same, 2.0 = double).
 #scale = 1.0
 
-# Override image height and/or width. Mutually exclusive with scale.
+# Resize to specified height, width, or both. Mutually exclusive with scale.
 #height = 0
 #width = 0
 
-# Method to use for image scaling (nearest, linear, cubic, area, lanczos4).
+# Method to use for scaling (nearest, linear, cubic, area, lanczos4).
 #scale-method = linear
 
+# Use separate threads for encoding and disk IO. Can improve performance.
+#threading = yes
+
 
 [export-html]
 # Filename format of created HTML file. Can use $VIDEO_NAME in the name.

diff --git a/scenedetect/__init__.py b/scenedetect/__init__.py
@@ -22,43 +22,37 @@
 # need to support both opencv-python and opencv-python-headless. Include some additional
 # context with the exception if this is the case.
 try:
-    import cv2
+    import cv2 as _
 except ModuleNotFoundError as ex:
     raise ModuleNotFoundError(
         "OpenCV could not be found, try installing opencv-python:\n\npip install opencv-python",
         name="cv2",
     ) from ex
-import numpy as np
 
-from scenedetect.backends import (
-    AVAILABLE_BACKENDS,
-    VideoCaptureAdapter,
-    VideoStreamAv,
-    VideoStreamCv2,
-    VideoStreamMoviePy,
-)
+# Commonly used classes/functions exported under the `scenedetect` namespace for brevity.
+# Note that order of importants is important!
+from scenedetect.platform import init_logger  # noqa: I001
+from scenedetect.frame_timecode import FrameTimecode
+from scenedetect.video_stream import VideoStream, VideoOpenFailure
+from scenedetect.video_splitter import split_video_ffmpeg, split_video_mkvmerge
+from scenedetect.scene_detector import SceneDetector
 from scenedetect.detectors import (
-    AdaptiveDetector,
     ContentDetector,
-    HashDetector,
-    HistogramDetector,
+    AdaptiveDetector,
     ThresholdDetector,
+    HistogramDetector,
+    HashDetector,
 )
-from scenedetect.frame_timecode import FrameTimecode
-
-# Commonly used classes/functions exported under the `scenedetect` namespace for brevity.
-from scenedetect.platform import (  # noqa: I001
-    get_and_create_path,
-    get_cv2_imwrite_params,
-    init_logger,
-    tqdm,
+from scenedetect.backends import (
+    AVAILABLE_BACKENDS,
+    VideoStreamCv2,
+    VideoStreamAv,
+    VideoStreamMoviePy,
+    VideoCaptureAdapter,
 )
-from scenedetect.scene_detector import SceneDetector
-from scenedetect.scene_manager import Interpolation, SceneList, SceneManager, save_images
-from scenedetect.stats_manager import StatsFileCorrupt, StatsManager
+from scenedetect.stats_manager import StatsManager, StatsFileCorrupt
+from scenedetect.scene_manager import SceneManager, save_images, SceneList, CutList, Interpolation
 from scenedetect.video_manager import VideoManager  # [DEPRECATED] DO NOT USE.
-from scenedetect.video_splitter import split_video_ffmpeg, split_video_mkvmerge
-from scenedetect.video_stream import VideoOpenFailure, VideoStream
 
 # Used for module identification and when printing version & about info
 # (e.g. calling `scenedetect version` or `scenedetect about`).

diff --git a/scenedetect/_cli/__init__.py b/scenedetect/_cli/__init__.py
@@ -1494,6 +1494,7 @@ def save_images_command(
         "output_dir": output,
         "scale": scale,
         "show_progress": not ctx.quiet_mode,
+        "threading": ctx.config.get_value("save-images", "threading"),
         "width": width,
     }
     ctx.add_command(cli_commands.save_images, save_images_args)

diff --git a/scenedetect/_cli/commands.py b/scenedetect/_cli/commands.py
@@ -177,6 +177,7 @@ def save_images(
     height: int,
     width: int,
     interpolation: Interpolation,
+    threading: bool,
 ):
     """Handles the `save-images` command."""
     del cuts  # save-images only uses scenes.
@@ -195,6 +196,7 @@ def save_images(
         height=height,
         width=width,
         interpolation=interpolation,
+        threading=threading,
     )
     # Save the result for use by `export-html` if required.
     context.save_images_result = (images, output_dir)

diff --git a/scenedetect/_cli/config.py b/scenedetect/_cli/config.py
@@ -361,6 +361,7 @@ def format(self, timecode: FrameTimecode) -> str:
         "quality": RangeValue(_PLACEHOLDER, min_val=0, max_val=100),
         "scale": 1.0,
         "scale-method": Interpolation.LINEAR,
+        "threading": True,
         "width": 0,
     },
     "save-qp": {

diff --git a/scenedetect/scene_manager.py b/scenedetect/scene_manager.py
@@ -213,7 +213,7 @@ def get_scenes_from_cuts(
     return scene_list
 
 
-# TODO(v1.0): Move post-processing functionality into separate submodule.
+# TODO(#463): Move post-processing functionality into separate submodule.
 
 
 def write_scene_list(

diff --git a/tests/test_scene_manager.py b/tests/test_scene_manager.py
@@ -18,6 +18,7 @@
 import glob
 import os
 import os.path
+from pathlib import Path
 from typing import List
 
 from scenedetect.backends.opencv import VideoStreamCv2
@@ -84,7 +85,7 @@ def test_get_scene_list_start_in_scene(test_video_file):
     assert scene_list[0][1] == end_time
 
 
-def test_save_images(test_video_file):
+def test_save_images(test_video_file, tmp_path: Path):
     """Test scenedetect.scene_manager.save_images function."""
     video = VideoStreamCv2(test_video_file)
     sm = SceneManager()
@@ -97,66 +98,101 @@ def test_save_images(test_video_file):
         "$TIMESTAMP_MS.$TIMECODE"
     )
 
-    try:
-        video_fps = video.frame_rate
-        scene_list = [
-            (FrameTimecode(start, video_fps), FrameTimecode(end, video_fps))
-            for start, end in [(0, 100), (200, 300), (300, 400)]
-        ]
+    video_fps = video.frame_rate
+    scene_list = [
+        (FrameTimecode(start, video_fps), FrameTimecode(end, video_fps))
+        for start, end in [(0, 100), (200, 300), (300, 400)]
+    ]
+
+    image_filenames = save_images(
+        scene_list=scene_list,
+        output_dir=tmp_path,
+        video=video,
+        num_images=3,
+        image_extension="jpg",
+        image_name_template=image_name_template,
+        threading=False,
+    )
 
-        image_filenames = save_images(
-            scene_list=scene_list,
-            video=video,
-            num_images=3,
-            image_extension="jpg",
-            image_name_template=image_name_template,
-        )
+    # Ensure images got created, and the proper number got created.
+    total_images = 0
+    for scene_number in image_filenames:
+        for path in image_filenames[scene_number]:
+            assert tmp_path.joinpath(path).exists(), f"expected {path} to exist"
+            total_images += 1
 
-        # Ensure images got created, and the proper number got created.
-        total_images = 0
-        for scene_number in image_filenames:
-            for path in image_filenames[scene_number]:
-                assert os.path.exists(path), f"expected {path} to exist"
-                total_images += 1
+    assert total_images == len([path for path in tmp_path.glob(image_name_glob)])
 
-        assert total_images == len(glob.glob(image_name_glob))
 
-    finally:
-        for path in glob.glob(image_name_glob):
-            os.remove(path)
+def test_save_images_singlethreaded(test_video_file, tmp_path: Path):
+    """Test scenedetect.scene_manager.save_images function."""
+    video = VideoStreamCv2(test_video_file)
+    sm = SceneManager()
+    sm.add_detector(ContentDetector())
+
+    image_name_glob = "scenedetect.tempfile.*.jpg"
+    image_name_template = (
+        "scenedetect.tempfile."
+        "$SCENE_NUMBER.$IMAGE_NUMBER.$FRAME_NUMBER."
+        "$TIMESTAMP_MS.$TIMECODE"
+    )
+
+    video_fps = video.frame_rate
+    scene_list = [
+        (FrameTimecode(start, video_fps), FrameTimecode(end, video_fps))
+        for start, end in [(0, 100), (200, 300), (300, 400)]
+    ]
+
+    image_filenames = save_images(
+        scene_list=scene_list,
+        output_dir=tmp_path,
+        video=video,
+        num_images=3,
+        image_extension="jpg",
+        image_name_template=image_name_template,
+        threading=True,
+    )
+
+    # Ensure images got created, and the proper number got created.
+    total_images = 0
+    for scene_number in image_filenames:
+        for path in image_filenames[scene_number]:
+            assert tmp_path.joinpath(path).exists(), f"expected {path} to exist"
+            total_images += 1
+
+    assert total_images == len([path for path in tmp_path.glob(image_name_glob)])
 
 
 # TODO: Test other functionality against zero width scenes.
-def test_save_images_zero_width_scene(test_video_file):
+def test_save_images_zero_width_scene(test_video_file, tmp_path: Path):
     """Test scenedetect.scene_manager.save_images guards against zero width scenes."""
     video = VideoStreamCv2(test_video_file)
     image_name_glob = "scenedetect.tempfile.*.jpg"
     image_name_template = "scenedetect.tempfile.$SCENE_NUMBER.$IMAGE_NUMBER"
-    try:
-        video_fps = video.frame_rate
-        scene_list = [
-            (FrameTimecode(start, video_fps), FrameTimecode(end, video_fps))
-            for start, end in [(0, 0), (1, 1), (2, 3)]
-        ]
-        NUM_IMAGES = 10
-        image_filenames = save_images(
-            scene_list=scene_list,
-            video=video,
-            num_images=10,
-            image_extension="jpg",
-            image_name_template=image_name_template,
-        )
-        assert len(image_filenames) == 3
-        assert all(len(image_filenames[scene]) == NUM_IMAGES for scene in image_filenames)
-        total_images = 0
-        for scene_number in image_filenames:
-            for path in image_filenames[scene_number]:
-                assert os.path.exists(path), f"expected {path} to exist"
-                total_images += 1
-        assert total_images == len(glob.glob(image_name_glob))
-    finally:
-        for path in glob.glob(image_name_glob):
-            os.remove(path)
+
+    video_fps = video.frame_rate
+    scene_list = [
+        (FrameTimecode(start, video_fps), FrameTimecode(end, video_fps))
+        for start, end in [(0, 0), (1, 1), (2, 3)]
+    ]
+    NUM_IMAGES = 10
+    image_filenames = save_images(
+        scene_list=scene_list,
+        output_dir=tmp_path,
+        video=video,
+        num_images=10,
+        image_extension="jpg",
+        image_name_template=image_name_template,
+    )
+    assert len(image_filenames) == 3
+    assert all(len(image_filenames[scene]) == NUM_IMAGES for scene in image_filenames)
+    total_images = 0
+    for scene_number in image_filenames:
+        for path in image_filenames[scene_number]:
+            assert tmp_path.joinpath(path).exists(), f"expected {path} to exist"
+            total_images += 1
+
+    assert total_images == len([path for path in tmp_path.glob(image_name_glob)])
 
 
 # TODO: This would be more readable if the callbacks were defined within the test case, e.g.

diff --git a/website/pages/changelog.md b/website/pages/changelog.md
@@ -588,15 +588,18 @@ Development
  - [feature] Add ability to configure CSV separators for rows/columns in config file [#423](https://github.com/Breakthrough/PySceneDetect/issues/423)
  - [feature] Add new `--show` flag to `export-html` command to launch browser after processing [#442](https://github.com/Breakthrough/PySceneDetect/issues/442)
  - [general] Timecodes of the form `MM:SS[.nnn]` are now processed correctly [#443](https://github.com/Breakthrough/PySceneDetect/issues/443)
- - [bugfix] Fix `save-images`/`save_images()` not working correctly with UTF-8 paths [#450](https://github.com/Breakthrough/PySceneDetect/issues/455)
+ - [bugfix] Fix `save-images`/`save_images()` not working correctly with UTF-8 paths [#450](https://github.com/Breakthrough/PySceneDetect/issues/450)
+ - [improvement] Add new `threading` option to `save-images`/`save_images()` [#456](https://github.com/Breakthrough/PySceneDetect/issues/456)
+    - Enabled by default, offloads image encoding and disk IO to separate threads
+    - Improves performance by up to 50% in some cases
  - [bugfix] Fix crash when using `save-images`/`save_images()` with OpenCV backend [#455](https://github.com/Breakthrough/PySceneDetect/issues/455)
  - [bugfix] Fix new detectors not working with `default-detector` config option
  - [improvement] The `export-html` command now implicitly invokes `save-images` with default parameters
-     - The output of the `export-html` command will always use the result of the `save-images` command that *precedes* it
+    - The output of the `export-html` command will always use the result of the `save-images` command that *precedes* it
  - [general] Updates to Windows distributions:
     - The MoviePy backend is now included with Windows distributions
     - Bundled Python interpreter is now Python 3.13
     - Updated PyAV 10 -> 13.1.0 and OpenCV 4.10.0.82 -> 4.10.0.84
  - [improvement] `save_to_csv` now works with paths from `pathlib`
  - [api] The `save_to_csv` function now works correctly with paths from the `pathlib` module
- - [api] Add `col_separator` and `row_separator` args to `write_scene_list` function in `scenedetect.scene_manager`
+ - [api] Add `col_separator` and `row_separator` args to `write_scene_list` function in `scenedetect.scene_manager`