코일100

  • 설명 :

데이터 세트에는 100개 개체의 7200개 컬러 이미지(개체당 72개 이미지)가 포함되어 있습니다. 물체는 매우 다양한 복잡한 기하학적 특성과 반사 특성을 가지고 있습니다. 개체는 검정색 배경에 대해 전동식 턴테이블에 배치되었습니다. 턴테이블을 360도 회전시켜 고정된 컬러 카메라에 따라 물체의 자세를 변경했습니다. 개체의 이미지는 5도의 포즈 간격으로 촬영되었습니다. 이는 개체당 72개의 포즈에 해당합니다.

나뉘다
'train' 7,200
  • 기능 구조 :
FeaturesDict({
    'angle': int64,
    'angle_label': ClassLabel(shape=(), dtype=int64, num_classes=72),
    'image': Image(shape=(128, 128, 3), dtype=uint8),
    'object_id': ClassLabel(shape=(), dtype=int64, num_classes=100),
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
각도 텐서 int64
angle_label 클래스 레이블 int64
영상 영상 (128, 128, 3) uint8
object_id 클래스 레이블 int64

심상

  • 인용 :
@article{nene1996columbia,
  title={Columbia object image library (coil-20)},
  author={Nene, Sameer A and Nayar, Shree K and Murase, Hiroshi and others},
  year={1996},
  publisher={Technical report CUCS-005-96}
}