open_images_v4

Gambar Terbuka adalah kumpulan data ~9 juta gambar yang telah dianotasi dengan label tingkat gambar dan kotak pembatas objek.

Set pelatihan V4 berisi 14,6 juta kotak pembatas untuk 600 kelas objek pada 1,74 juta gambar, menjadikannya kumpulan data terbesar yang ada dengan anotasi lokasi objek. Kotak-kotak tersebut sebagian besar telah digambar secara manual oleh anotator profesional untuk memastikan akurasi dan konsistensi. Gambar-gambarnya sangat beragam dan seringkali berisi pemandangan kompleks dengan beberapa objek (rata-rata 8,4 per gambar). Selain itu, kumpulan data dianotasi dengan label level gambar yang mencakup ribuan kelas.

Membelah Contoh
'test' 125.436
'train' 1.743.042
'validation' 41.620
  • Struktur fitur :
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),
    }),
})
  • Dokumentasi fitur :
Fitur Kelas Membentuk Dtype Keterangan
fiturDict
bobjects Urutan
bobjects/bbox Fitur BBox (4,) float32
bobjects/is_depiction Tensor int8
bobjects/is_group_of Tensor int8
bobjects/is_inside Tensor int8
bobjects/is_occluded Tensor int8
bobjects/is_truncated Tensor int8
bobjects/label LabelKelas int64
bobjects/sumber LabelKelas int64
gambar Gambar (Tidak ada, Tidak ada, 3) uint8
gambar/nama file Teks rangkaian
objek Urutan
benda/kepercayaan Tensor int32
benda/label LabelKelas int64
benda/sumber LabelKelas int64
objek_dapat dilatih Urutan
objek_dapat dilatih/percaya diri Tensor int32
objek_dapat dilatih/label LabelKelas int64
objek_dapat dilatih/sumber LabelKelas 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/asli (konfigurasi default)

  • Deskripsi konfigurasi : Gambar dengan resolusi dan kualitas aslinya.

  • Ukuran dataset : 562.42 GiB

  • Gambar ( tfds.show_examples ):

Visualisasi

buka_gambar_v4/300k

  • Deskripsi konfigurasi : Gambar memiliki sekitar 300.000 piksel, dengan kualitas 72 JPEG.

  • Ukuran dataset : 81.92 GiB

  • Gambar ( tfds.show_examples ):

Visualisasi

buka_gambar_v4/200k

  • Deskripsi konfigurasi : Gambar memiliki sekitar 200.000 piksel, dengan kualitas 72 JPEG.

  • Ukuran dataset : 60.70 GiB

  • Gambar ( tfds.show_examples ):

Visualisasi