Description #

Image safety (NSFW) detector—supports multi-source input and batch inference

Key features #

• Uses a pre-trained image classification model for NSFW probability detection and outputs the NSFW confidence score for each image
• Supports multiple input sources: URL address (image_url), Base64 encoding (image_base64), binary stream (image_binary)
• Batch processing: Configure batch_size for batch inference to improve throughput performance

Application scenarios #

• Content safety review: image review process for e-commerce and social platforms
• Pre-filtering in production pipelines: perform safety screening before image processing or generation tasks

Cautions #

• Only outputs numeric NSFW confidence (0~1) and does not save images. To persist results, implement persistence in the upstream or downstream pipeline.

Operator parameters #

Input #

Output #

Returns an array containing the detection results, where each element is the NSFW (Not Safe for Work) confidence score (floating-point value) for the corresponding image. If detection fails, the element is None.

• nsfw_detect (float | None): NSFW confidence score. The closer this value is to 1.0, the higher the likelihood that the image is determined to contain inappropriate content.

Parameters #

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

Examples #

The following code demonstrates how to use Daft and LAS UDF to perform NSFW detection on images.

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.image.image_nsfw_detect import ImageNsfwDetect
from daft.las.functions.udf import las_udf

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.%s".format(),
            )
            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)

    tos_dir_url = os.getenv("TOS_DIR_URL", "las-cn-beijing-public-online.tos-cn-beijing.volces.com")
    samples = {
        "image": [
            f"https://{tos_dir_url}/public/shared_image_dataset/cat_ip_adapter.jpeg"
        ]
    }

    image_src_type = "image_url"
    model_path = os.getenv("MODEL_PATH", "/opt/las/models")
    model_name = "Falconsai/nsfw_image_detection"
    rank = 0
    num_gpus = 0
    batch_size = 1

    ds = daft.from_pydict(samples)
    ds = ds.with_column(
        "nsfw_detect",
        las_udf(
            ImageNsfwDetect,
            construct_args={
                "image_src_type": image_src_type,
                "batch_size": batch_size,
                "model_path": model_path,
                "model_name": model_name,
                "rank": rank,
            },
            num_gpus=num_gpus,
            batch_size=1,
        )(col("image")),
    )

    ds.show()

    # ╭────────────────────────────────┬────────────────────────╮
    # │ image                          ┆ nsfw_detect            │
    # │ ---                            ┆ ---                    │
    # │ Utf8                           ┆ Float64                │
    # ╞════════════════════════════════╪════════════════════════╡
    # │ https://las-cn-beijing-public… ┆ 0.000114               │
    # ╰────────────────────────────────┴────────────────────────╯