Description #

Image hash calculation operator: Supports three input formats—URL, Base64, and binary—and outputs both hexadecimal and binary hashes in a unified manner.

Key features #

• Supports five hash methods: ahash, dhash, phash, whash, md5
• Selectable input types: image_url, image_base64, image_binary
• Supports batch processing

Applicable scenarios #

• Image deduplication and recall of near-duplicate images
• File consistency verification
• Multi-format hash value output

Caution #

• The input data type must match the image_src_type parameter
• The output of md5 and perceptual hash will be standardized to 64-bit binary and 16-digit hexadecimal
• The operator depends on imagededup.methods; if the dependency is unavailable, an empty string will be returned
• Images are uniformly read and converted to RGB format using Pillow

Supported hash methods #

• ahash
• dhash
• phash
• whash
• md5

Operator parameters #

Input #

Output #

A structured result array, where each element contains the following fields:

• hash_hex (str): The hexadecimal hash value of the image. For the md5 method, this value is 16 characters (the first 64 bits of the full digest).
• hash_bin (str): The binary hash value of the image. For the md5 method, this value is 64 bits.

Parameters #

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

Examples #

The following code demonstrates how to use daft to invoke the ImageHash operator to compute the perceptual hash value of an image.

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.image.image_hash import ImageHash
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"
        ]
    }

    ds = daft.from_pydict(samples)
    ds = ds.with_column(
        "image_hash",
        las_udf(
            ImageHash,
            construct_args={
                "image_src_type": "image_url",
                "method": "phash",
            },
            num_gpus=0,
            batch_size=1,
            concurrency=1,
        )(col("image")),
    )

    ds.show()

    # ╭────────────────────────────────┬────────────────────────────────────────╮
    # │ image                          ┆ image_hash                             │
    # │ ---                            ┆ ---                                    │
    # │ Utf8                           ┆ Struct[hash_hex: Utf8, hash_bin: Utf8] │
    # ╞════════════════════════════════╪════════════════════════════════════════╡
    # │ https://las-cn-beijing-public… ┆ {hash_hex: 8d3986a636e768ad,           │
    # │                                ┆ …                                      │
    # ╰────────────────────────────────┴────────────────────────────────────────╯