dcase23-task2-enriched.py

import os
import json
import datasets
import datasets.info
import pandas as pd
import numpy as np
import tempfile
import requests
import io
from pathlib import Path
from datasets import load_dataset
from typing import Iterable, Dict, Optional, Union, List


_CITATION = """\
@dataset{kota_dohi_2023_7882613,
  author       = {Kota Dohi and
                  Keisuke Imoto and
                  Noboru Harada and
                  Daisuke Niizumi and
                  Yuma Koizumi and
                  Tomoya Nishida and
                  Harsh Purohit and
                  Takashi Endo and
                  Yohei Kawaguchi},
  title        = {DCASE 2023 Challenge Task 2 Development Dataset},
  month        = mar,
  year         = 2023,
  publisher    = {Zenodo},
  version      = {3.0},
  doi          = {10.5281/zenodo.7882613},
  url          = {https://doi.org/10.5281/zenodo.7882613}
}
"""
_LICENSE = "Creative Commons Attribution 4.0 International Public License"

_METADATA_REG = r"attributes_\d+.csv"

_NUM_TARGETS = 2
_NUM_CLASSES = 14

_TARGET_NAMES = ["normal", "anomaly"]
_CLASS_NAMES = ["gearbox", "fan", "bearing", "slider", "ToyCar", "ToyTrain", "valve", "bandsaw", "grinder", "shaker", "ToyDrone", "ToyNscale", "ToyTank", "Vacuum"]

_HOMEPAGE = {
    "dev": "https://zenodo.org/record/7690157",
    "add": "",
    "eval": "",
}

DATA_URLS = {
    "dev": {
        "train": "data/dev_train.tar.gz",
        "test": "data/dev_test.tar.gz",
        "metadata": "data/dev_metadata.csv",
    },
    "add":  {
        "train": "data/add_train.tar.gz",
        "metadata": "data/add_metadata.csv",
    },
    "eval": {
        "test": "data/eval_test.tar.gz",
        "metadata": None,
    },
}

EMBEDDING_URLS = {
    "dev": {
        "embeddings_ast-finetuned-audioset-10-10-0.4593": {
            "train": "data/MIT_ast-finetuned-audioset-10-10-0.4593-embeddings_dev_train.npz",
            "test": "data/MIT_ast-finetuned-audioset-10-10-0.4593-embeddings_dev_test.npz",
            "size": (1, 768),
            "dtype": "float32",
        },
    },
    "add":  {
        "embeddings_ast-finetuned-audioset-10-10-0.4593": {
            "train": "data/MIT_ast-finetuned-audioset-10-10-0.4593-embeddings_add_train.npz",
            "size": (1, 768),
            "dtype": "float32",
        },
    },
    "eval": {
        "embeddings_ast-finetuned-audioset-10-10-0.4593": {
            "test": "data/MIT_ast-finetuned-audioset-10-10-0.4593-embeddings_eval_test.npz",
            "size": (1, 768),
            "dtype": "float32",
        },
    },
}

STATS = {
    "name": "Enriched Dataset of 'DCASE 2023 Challenge Task 2'",
    "configs": {
        'dev': {
            'date': "Mar 1, 2023",
            'version': "3.0.0",
            'homepage': "https://zenodo.org/record/7882613",
            "splits": ["train", "test"],
        },
        'add': {
            'date': "Apr 15, 2023",
            'version': "1.0.0",
            'homepage': "https://zenodo.org/record/7830345",
            "splits": ["train"],
        },
        'eval': {
            'date': "May 1, 2023",
            'version': "1.0.0",
            'homepage': "https://zenodo.org/record/7860847",
            "splits": ["test"],
        },
    }
}

DATASET = {
    'dev': 'DCASE 2023 Challenge Task 2 Development Dataset',
    'add': 'DCASE 2023 Challenge Task 2 Additional Train Dataset',
    'eval': 'DCASE 2023 Challenge Task 2 Evaluation Dataset',
}


SPOTLIGHT_LAYOUTS = {
    "standard": {"orientation":"vertical","children":[{"kind":"split","weight":51.96463654223969,"orientation":"horizontal","children":[{"kind":"tab","weight":30,"children":[{"kind":"widget","name":"Table","type":"table","config":{"tableView":"full","visibleColumns":["class","class_name","config","d1p","d1v","d2p","d2v","d3p","d3v","file_path","label","section","split"],"sorting":None,"orderByRelevance":False}}]},{"kind":"tab","weight":33.970588235294116,"children":[{"kind":"widget","name":"Similarity Map (2)","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"label","sizeBy":None,"filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}}]},{"kind":"tab","weight":36.029411764705884,"children":[{"kind":"widget","name":"Similarity Map","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"class","sizeBy":None,"filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}},{"kind":"widget","name":"Scatter Plot","type":"scatterplot","config":{"xAxisColumn":None,"yAxisColumn":None,"colorBy":None,"sizeBy":None,"filter":False}},{"kind":"widget","name":"Histogram","type":"histogram","config":{"columnKey":None,"stackByColumnKey":None,"filter":False}}]}]},{"kind":"tab","weight":48.03536345776031,"children":[{"kind":"widget","name":"Inspector","type":"inspector","config":{"views":[{"view":"AudioView","columns":["path"],"name":"view","key":"43a5beff-9423-41c9-a5ba-285a7ece7a02"},{"view":"SpectrogramView","columns":["path"],"name":"view","key":"5f035027-dd02-4587-ba77-defdf823c124"}],"visibleColumns":4}}]}]},
    "simple": {"orientation":"vertical","children":[{"kind":"split","weight":60.575296108291035,"orientation":"horizontal","children":[{"kind":"tab","weight":31.52260461369049,"children":[{"kind":"widget","name":"Table","type":"table","config":{"tableView":"filtered","visibleColumns":["class","d1p","d1v","d2p","d2v","d3p","d3v","dev_train_lof_anomaly","dev_train_lof_anomaly_score","domain","label","section"],"sorting":None,"orderByRelevance":False}}]},{"kind":"tab","weight":33.869200490640154,"children":[{"kind":"widget","name":"Similarity map with AST-lof anomaly score","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"dev_train_lof_anomaly_score","sizeBy":"label","filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}}]},{"kind":"tab","weight":34.60819489566936,"children":[{"kind":"widget","name":"Similarity map with classes","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"class","sizeBy":None,"filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}},{"kind":"widget","name":"Scatter Plot","type":"scatterplot","config":{"xAxisColumn":None,"yAxisColumn":None,"colorBy":None,"sizeBy":None,"filter":False}},{"kind":"widget","name":"Histogram","type":"histogram","config":{"columnKey":"domain","stackByColumnKey":"prediction_correct_dcase2023_task2_baseline_ae","filter":False}}]}]},{"kind":"tab","weight":39.424703891708965,"children":[{"kind":"widget","name":"Inspector","type":"inspector","config":{"views":[{"view":"AudioView","columns":["path"],"name":"view","key":"dea9a175-9582-412e-9f49-be729e8838fb"},{"view":"SpectrogramView","columns":["path"],"name":"view","key":"676bd937-226b-4632-ae2d-ec8bc37bcc5d"},{"view":"ScalarView","columns":["label"],"name":"view","key":"dbfcc0b1-9e96-4d31-8856-f0bd7f0b8144"},{"view":"ScalarView","columns":["domain"],"name":"view","key":"3e79654f-e017-402c-b136-6a13c4409ae4"}],"visibleColumns":4}}]}]},
    "extended": {"orientation":"vertical","children":[{"kind":"split","weight":54.145516074450086,"orientation":"horizontal","children":[{"kind":"tab","weight":31.52260461369049,"children":[{"kind":"widget","name":"Table","type":"table","config":{"tableView":"filtered","visibleColumns":["class","d1p","d1v","d2p","d2v","d3p","d3v","dev_train_lof_anomaly","dev_train_lof_anomaly_score","domain","label","section"],"sorting":None,"orderByRelevance":False}}]},{"kind":"tab","weight":33.869200490640154,"children":[{"kind":"widget","name":"Similarity map with AST-lof anomaly score","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"dev_train_lof_anomaly_score","sizeBy":"label","filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}}]},{"kind":"tab","weight":34.60819489566936,"children":[{"kind":"widget","name":"Similarity map with classes","type":"similaritymap","config":{"placeBy":None,"reductionMethod":None,"colorBy":"class","sizeBy":None,"filter":False,"umapNNeighbors":20,"umapMetric":None,"umapMinDist":0.15,"pcaNormalization":None,"umapMenuLocalGlobalBalance":None,"umapMenuIsAdvanced":False}},{"kind":"widget","name":"Scatter Plot","type":"scatterplot","config":{"xAxisColumn":None,"yAxisColumn":None,"colorBy":None,"sizeBy":None,"filter":False}}]}]},{"kind":"split","weight":45.854483925549914,"orientation":"horizontal","children":[{"kind":"tab","weight":58.581483486735245,"children":[{"kind":"widget","name":"Inspector","type":"inspector","config":{"views":[{"view":"AudioView","columns":["path"],"name":"view","key":"dea9a175-9582-412e-9f49-be729e8838fb"},{"view":"SpectrogramView","columns":["path"],"name":"view","key":"676bd937-226b-4632-ae2d-ec8bc37bcc5d"},{"view":"ScalarView","columns":["label"],"name":"view","key":"dbfcc0b1-9e96-4d31-8856-f0bd7f0b8144"},{"view":"ScalarView","columns":["domain"],"name":"view","key":"3e79654f-e017-402c-b136-6a13c4409ae4"}],"visibleColumns":4}}]},{"kind":"tab","weight":41.418516513264755,"children":[{"kind":"widget","name":"Histogram","type":"histogram","config":{"columnKey":"class","stackByColumnKey":"dev_train_lof_anomaly"}}]}]}]},
}

SPOTLIGHT_RENAME = {
    "audio": "original_audio",
    "path": "audio",
}


class DCASE2023Task2DatasetConfig(datasets.BuilderConfig):
    """BuilderConfig for DCASE2023Task2Dataset."""

    def __init__(self, name, version, **kwargs):
        self.release_date = kwargs.pop("release_date", None)
        self.homepage = kwargs.pop("homepage", None)
        self.data_urls = kwargs.pop("data_urls", None)
        self.embeddings_urls = kwargs.pop("embeddings_urls", None)
        self.splits = kwargs.pop("splits", None)
        self.rename = kwargs.pop("rename", None)
        self.layout = kwargs.pop("layout", None)
        description = (
            f"Dataset for the DCASE 2023 Challenge Task 2 'First-Shot Unsupervised Anomalous Sound Detection "
            f"for Machine Condition Monitoring'. released on {self.release_date}. Original data available under"
            f"{self.homepage}. "
            f"CONFIG: {name}."
        )
        super(DCASE2023Task2DatasetConfig, self).__init__(
            name=name,
            version=datasets.Version(version),
            description=description,
        )

    def to_spotlight(self, data: Union[pd.DataFrame, datasets.Dataset]) -> pd.DataFrame:

        def get_split(path: str) -> str:
            fn = os.path.basename(path)
            if "train" in fn:
                return "train"
            elif "test" in fn:
                return "test"
            else:
                raise NotImplementedError

        if type(data) == datasets.Dataset:
            # retrieve split
            df = data.to_pandas()
            df["split"] = data.split._name if "+" not in data.split._name else df["path"].map(get_split)
            df["config"] = data.config_name

            # get clearnames for classes
            class_names = data.features["class"].names
            df["class_name"] = df["class"].apply(lambda x: class_names[x])
        elif type(data) == pd.DataFrame:
            df = data
        else:
            raise TypeError("type(data) not in Union[pd.DataFrame, datasets.Dataset]")

        df["file_path"] = df["path"]
        df.rename(columns=self.rename, inplace=True)

        return df.copy()

    def get_layout(self, config: str = "standard") -> str:
        layout_json = tempfile.mktemp(".json")
        with open(layout_json, "w") as outfile:
            json.dump(self.layout[config], outfile)

        return layout_json


class DCASE2023Task2Dataset(datasets.GeneratorBasedBuilder):
    """Dataset for the DCASE 2023 Challenge Task 2 "First-Shot Unsupervised Anomalous Sound Detection
    for Machine Condition Monitoring"."""

    VERSION = datasets.Version("0.1.0")

    DEFAULT_CONFIG_NAME = "dev"

    BUILDER_CONFIGS = [
        DCASE2023Task2DatasetConfig(
            name=key,
            version=stats["version"],
            dataset=DATASET[key],
            homepage=_HOMEPAGE[key],
            data_urls=DATA_URLS[key],
            embeddings_urls=EMBEDDING_URLS[key],
            release_date=stats["date"],
            splits=stats["splits"],
            layout=SPOTLIGHT_LAYOUTS,
            rename=SPOTLIGHT_RENAME,
        )
        for key, stats in STATS["configs"].items()
    ]

    def _info(self):
        features = {
                    "audio": datasets.Audio(sampling_rate=16_000),
                    "path": datasets.Value("string"),
                    "section": datasets.Value("int64"),
                    "domain": datasets.ClassLabel(num_classes=2, names=["source", "target"]),
                    "label": datasets.ClassLabel(num_classes=_NUM_TARGETS, names=_TARGET_NAMES),
                    "class": datasets.ClassLabel(num_classes=_NUM_CLASSES, names=_CLASS_NAMES),
                    "d1p": datasets.Value("string"),
                    "d1v": datasets.Value("string"),
                    "d2p": datasets.Value("string"),
                    "d2v": datasets.Value("string"),
                    "d3p": datasets.Value("string"),
                    "d3v": datasets.Value("string"),
                    "dev_train_lof_anomaly": datasets.Value("int64"),
                    "dev_train_lof_anomaly_score": datasets.Value("float32"),
                    "add_train_lof_anomaly": datasets.Value("int64"),
                    "add_train_lof_anomaly_score": datasets.Value("float32"),
                }
        if self.config.embeddings_urls is not None:
            features.update({
                emb_name: [datasets.Value(emb["dtype"])] for emb_name, emb in self.config.embeddings_urls.items()
            })
        features = datasets.Features(features)

        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=self.config.description,
            features=features,
            supervised_keys=datasets.info.SupervisedKeysData("label"),
            homepage=self.config.homepage,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(
            self,
            dl_manager: datasets.DownloadManager
    ):
        """Returns SplitGenerators."""
        dl_manager.download_config.ignore_url_params = True
        audio_path = {}
        local_extracted_archive = {}
        split_type = {"train": datasets.Split.TRAIN, "test": datasets.Split.TEST}
        embeddings = {split: dict() for split in split_type}

        for split in split_type:
            if split in self.config.splits:
                audio_path[split] = dl_manager.download(self.config.data_urls[split])
                local_extracted_archive[split] = dl_manager.extract(
                    audio_path[split]) if not dl_manager.is_streaming else None
                if self.config.embeddings_urls is not None:
                    for emb_name, emb_data in self.config.embeddings_urls.items():
                        downloaded_embeddings = dl_manager.download(emb_data[split])
                        if dl_manager.is_streaming:
                            response = requests.get(downloaded_embeddings)
                            response.raise_for_status()
                            downloaded_embeddings = io.BytesIO(response.content)
                        npz_file = np.load(downloaded_embeddings, allow_pickle=True)
                        embeddings[split][emb_name] = npz_file["arr_0"].item()

        return [
            datasets.SplitGenerator(
                name=split_type[split],
                gen_kwargs={
                    "split": split,
                    "local_extracted_archive": local_extracted_archive[split],
                    "audio_files": dl_manager.iter_archive(audio_path[split]),
                    "embeddings": embeddings[split],
                    "metadata_file": dl_manager.download_and_extract(self.config.data_urls["metadata"]) if self.config.data_urls["metadata"] is not None else None,
                    "scores_file": dl_manager.download_and_extract("data/scores.csv"),
                    "is_streaming": dl_manager.is_streaming,
                },
            ) for split in split_type if split in self.config.splits
        ]

    def _generate_examples(
        self,
        split: str,
        local_extracted_archive: Union[Dict, List],
        audio_files: Optional[Iterable],
        embeddings: Optional[Dict],
        metadata_file: Optional[str],
        scores_file: Optional[str],
        is_streaming: Optional[bool],
    ):
        """Yields examples."""
        if metadata_file is not None:
            metadata = pd.read_csv(metadata_file)
        if scores_file is not None:
            scores = pd.read_csv(scores_file)
        data_fields = list(self._info().features.keys())

        id_ = 0
        for path, f in audio_files:
            lookup = Path(path).parent.name + "/" + Path(path).name
            if metadata_file is None or lookup in metadata["path"].values:
                path = os.path.join(local_extracted_archive, path) if local_extracted_archive else path
                if is_streaming:
                    audio = {"path": path, "bytes": f.read()}
                else:
                    audio = {"path": path, "bytes": None}
                result = {field: None for field in data_fields}
                if metadata_file is not None:
                    result.update(metadata[metadata["path"] == lookup].T.squeeze().to_dict())
                if scores is not None:
                    result.update(scores[scores["path"] == lookup].T.squeeze().to_dict())
                for emb_key in embeddings.keys():
                    result[emb_key] = np.asarray(embeddings[emb_key][lookup]).squeeze().tolist()
                result["path"] = path
                yield id_, {**result, "audio": audio}
                id_ += 1


if __name__ == "__main__":
    ds = load_dataset("dcase23-task2-enriched.py", "dev", split="train", streaming=True)