open_images_v4

Open Images는 이미지 수준 레이블 및 개체 경계 상자로 주석이 달린 ~9M 이미지의 데이터 세트입니다.

V4의 트레이닝 세트에는 174만 개의 이미지에 있는 600개의 개체 클래스에 대한 1460만 개의 경계 상자가 포함되어 있어 개체 위치 주석이 있는 기존 데이터 세트 중 가장 큽니다. 상자는 정확성과 일관성을 보장하기 위해 대부분 전문 주석가가 수동으로 그렸습니다. 이미지는 매우 다양하며 종종 여러 개체가 있는 복잡한 장면을 포함합니다(이미지당 평균 8.4개). 또한 데이터 세트에는 수천 개의 클래스에 걸친 이미지 수준 레이블이 주석으로 추가됩니다.

나뉘다
'test' 125,436
'train' 1,743,042
'validation' 41,620
  • 기능 구조 :
FeaturesDict({
    'bobjects': Sequence({
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'is_depiction': int8,
        'is_group_of': int8,
        'is_inside': int8,
        'is_occluded': int8,
        'is_truncated': int8,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=601),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
    'objects': Sequence({
        'confidence': int32,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=19995),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
    'objects_trainable': Sequence({
        'confidence': int32,
        'label': ClassLabel(shape=(), dtype=int64, num_classes=7186),
        'source': ClassLabel(shape=(), dtype=int64, num_classes=6),
    }),
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
객체 순서
객체/bbox BBox기능 (4,) float32
bobjects/is_depiction 텐서 int8
bobjects/is_group_of 텐서 int8
bobjects/is_inside 텐서 int8
bobjects/is_occluded 텐서 int8
bobjects/is_truncated 텐서 int8
개체/라벨 클래스 레이블 int64
개체/소스 클래스 레이블 int64
영상 영상 (없음, 없음, 3) uint8
이미지/파일 이름 텍스트
사물 순서
객체/자신감 텐서 int32
개체/레이블 클래스 레이블 int64
개체/소스 클래스 레이블 int64
objects_trainable 순서
objects_trainable/신뢰 텐서 int32
objects_trainable/레이블 클래스 레이블 int64
objects_trainable/소스 클래스 레이블 int64
@article{OpenImages,
  author = {Alina Kuznetsova and
            Hassan Rom and
            Neil Alldrin and
            Jasper Uijlings and
            Ivan Krasin and
            Jordi Pont-Tuset and
            Shahab Kamali and
            Stefan Popov and
            Matteo Malloci and
            Tom Duerig and
            Vittorio Ferrari},
  title = {The Open Images Dataset V4: Unified image classification,
           object detection, and visual relationship detection at scale},
  year = {2018},
  journal = {arXiv:1811.00982}
}
@article{OpenImages2,
  author = {Krasin, Ivan and
            Duerig, Tom and
            Alldrin, Neil and
            Ferrari, Vittorio
            and Abu-El-Haija, Sami and
            Kuznetsova, Alina and
            Rom, Hassan and
            Uijlings, Jasper and
            Popov, Stefan and
            Kamali, Shahab and
            Malloci, Matteo and
            Pont-Tuset, Jordi and
            Veit, Andreas and
            Belongie, Serge and
            Gomes, Victor and
            Gupta, Abhinav and
            Sun, Chen and
            Chechik, Gal and
            Cai, David and
            Feng, Zheyun and
            Narayanan, Dhyanesh and
            Murphy, Kevin},
  title = {OpenImages: A public dataset for large-scale multi-label and
           multi-class image classification.},
  journal = {Dataset available from
             https://storage.googleapis.com/openimages/web/index.html},
  year={2017}
}

open_images_v4/original(기본 구성)

  • 구성 설명 : 원래 해상도 및 품질의 이미지입니다.

  • 데이터세트 크기 : 562.42 GiB

  • 그림 ( tfds.show_examples ):

심상

open_images_v4/300k

  • 구성 설명 : 이미지에는 72 JPEG 품질의 약 300,000픽셀이 있습니다.

  • 데이터세트 크기 : 81.92 GiB

  • 그림 ( tfds.show_examples ):

심상

open_images_v4/200k

  • 구성 설명 : 이미지에는 72 JPEG 품질의 약 200,000픽셀이 있습니다.

  • 데이터세트 크기 : 60.70 GiB

  • 그림 ( tfds.show_examples ):

심상