Operator ID: daft.las.functions.video.video_sharpness.VideoSharpness

Description #

Video sharpness calculation processor that supports multiple sharpness evaluation methods and frame extraction strategies.

Key features #

• Video frame extraction:
  • Frame extraction strategy based on FPS, customizable number of frames extracted per second
  • Limits the maximum number of extracted frames to prevent infinite loops when processing long videos
  • Supports dual frame extraction schemes using ffmpeg and cv2 to improve stability
• Multiple sharpness evaluation methods:
  • Laplacian variance method (laplacian) – edge detection based on second-order derivatives, simple and fast calculation
  • Tenengrad method (tenengrad) – edge strength based on Sobel gradients, sensitive to noise
  • Brenner method (brenner) – simple calculation based on the grayscale difference of adjacent pixels, fastest speed
  • FFT high-frequency energy method (fft_highfreq) – sharpness evaluation based on frequency domain analysis, more sensitive to blurring
• Multi-pooling functionality:
  • Returns results for all pooling methods (mean, median, maximum, minimum, standard deviation)
  • Automatically handles NaN values and exceptions
• Supports multiple input formats:
  • URL address
  • Binary stream

Format support: #

• Input: Mainstream video formats such as MP4, AVI, MOV

Operator parameters #

Input #

Output #

Array of structs containing sharpness results for all pooling methods. Each element includes:

• mean: Mean sharpness
• median: Median sharpness
• max: Maximum sharpness
• min: Minimum sharpness
• std: Standard deviation of sharpness

Parameters #

If a parameter does not have a default value, it is required

Examples #

The following code demonstrates how to use Daft (for distributed scenarios) to run the operator for video sharpness calculation.

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.udf import las_udf
from daft.las.functions.video.video_sharpness import VideoSharpness

if __name__ == "__main__":
    if os.getenv("DAFT_RUNNER", "native") == "ray":
        import logging

        import ray

        def configure_logging():
            logging.basicConfig(
                level=logging.INFO,
                format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
                datefmt="%Y-%m-%d %H:%M:%S",
            )
            logging.getLogger("tracing.span").setLevel(logging.WARNING)
            logging.getLogger("daft_io.stats").setLevel(logging.WARNING)
            logging.getLogger("DaftStatisticsManager").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaScheduler").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaDispatcher").setLevel(logging.WARNING)

        ray.init(dashboard_host="0.0.0.0", runtime_env={"worker_process_setup_hook": configure_logging})
        daft.set_runner_ray()
    daft.set_execution_config(actor_udf_ready_timeout=600)
    daft.set_execution_config(min_cpu_per_task=0)

    # Use environment variables to construct the URL
    tos_dir_url = os.getenv("TOS_DIR_URL", "las-cn-beijing-public-online.tos-cn-beijing.volces.com")
    samples = {
        "video_path": [
            f"https://{tos_dir_url}/public/shared_video_dataset/sample.mp4"
        ]
    }

    ds = daft.from_pydict(samples)

    sharpness_calculator = las_udf(
        VideoSharpness,
        construct_args={
            "fps": 2.0,
            "method": "laplacian",
            "max_frames": 100,
        },
        num_gpus=1,
        concurrency=1,
        batch_size=1,
    )

    # Use Daft for distributed processing
    ds = ds.with_column("video_sharpness", sharpness_calculator(col("video_path")))

    ds.show()
    # ╭────────────────────────────────┬──────────────────────────────────╮
    # │ video_path                     ┆ video_sharpness                  │
    # │ ---                            ┆ ---                              │
    # │ Utf8                           ┆ Struct[mean: Float64, median: F… │
    # ╞════════════════════════════════╪══════════════════════════════════╡
    # │ https://las-cn-beijing-publi-… ┆ {mean: 1234.56, median: 1200.00… │
    # ╰────────────────────────────────┴──────────────────────────────────╯