옥스포드_꽃102

  • 설명 :

Oxford Flowers 102 데이터 세트는 영국에서 일반적으로 발생하는 102개의 꽃 범주와 일치합니다. 각 클래스는 40~258개의 이미지로 구성됩니다. 이미지에는 대규모, 포즈 및 조명 변형이 있습니다. 또한 범주 내에서 큰 변형이 있는 범주와 여러 매우 유사한 범주가 있습니다.

데이터 세트는 훈련 세트, 검증 세트 및 테스트 세트로 나뉩니다. 트레이닝 세트와 검증 세트는 각각 클래스당 10개의 이미지로 구성됩니다(각각 총 1020개의 이미지). 테스트 세트는 나머지 6149개 이미지(클래스당 최소 20개)로 구성됩니다.

나뉘다
'test' 6,149
'train' 1,020
'validation' 1,020
  • 기능 구조 :
FeaturesDict({
    'file_name': Text(shape=(), dtype=string),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=102),
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
파일 이름 텍스트
영상 영상 (없음, 없음, 3) uint8
상표 클래스 레이블 int64

심상

  • 인용 :
@InProceedings{Nilsback08,
   author = "Nilsback, M-E. and Zisserman, A.",
   title = "Automated Flower Classification over a Large Number of Classes",
   booktitle = "Proceedings of the Indian Conference on Computer Vision, Graphics and Image Processing",
   year = "2008",
   month = "Dec"
}