bigearthnet

ה-BigEarthNet הוא ארכיון חדש בקנה מידה גדול של Sentinel-2, המורכב מ-590,326 תיקוני תמונה של Sentinel-2. גודל תיקון התמונה על הקרקע הוא 1.2 על 1.2 ק"מ עם גודל תמונה משתנה בהתאם לרזולוציית הערוץ. זהו מערך נתונים רב תוויות עם 43 תוויות לא מאוזנות.

כדי לבנות את BigEarthNet, נבחרו בתחילה 125 אריחי Sentinel-2 שנרכשו בין יוני 2017 למאי 2018 ב-10 המדינות (אוסטריה, בלגיה, פינלנד, אירלנד, קוסובו, ליטא, לוקסמבורג, פורטוגל, סרביה, שוויץ). כל האריחים תוקנו אטמוספרית על ידי כלי היצירה והעיצוב של המוצר Sentinel-2 Level 2A (sen2cor). לאחר מכן, הם חולקו ל-590,326 תיקוני תמונה שאינם חופפים. כל תיקון תמונה סומן על ידי מחלקות כיסוי הקרקע המרובות (כלומר, תוויות מרובות) שסופקו ממסד הנתונים של CORINE Land Cover של שנת 2018 (CLC 2018).

פסים ורזולוציית פיקסלים במטרים:

  • B01: אירוסול חופי; 60 מ'
  • B02: כחול; 10 מ'
  • B03: ירוק; 10 מ'
  • B04: אדום; 10 מ'
  • B05: קצה אדום של צמחיה; 20 מ'
  • B06: קצה אדום של צמחייה; 20 מ'
  • B07: קצה אדום של צמחיה; 20 מ'
  • B08: NIR; 10 מ'
  • B09: אדי מים; 60 מ'
  • B11: SWIR; 20 מ'
  • B12: SWIR; 20 מ'
  • B8A: NIR צר; 20 מ'

רישיון: הסכם רישיון נתונים קהילתי - מתיר, גרסה 1.0.

כתובת אתר: http://bigearth.net/

לְפַצֵל דוגמאות
'train' 590,326
  • ציטוט :
@article{Sumbul2019BigEarthNetAL,
  title={BigEarthNet: A Large-Scale Benchmark Archive For Remote Sensing Image Understanding},
  author={Gencer Sumbul and Marcela Charfuelan and Beg{"u}m Demir and Volker Markl},
  journal={CoRR},
  year={2019},
  volume={abs/1902.06148}
}

bigearthnet/rgb (תצורת ברירת המחדל)

  • תיאור תצורה : ערוצי Sentinel-2 RGB

  • גודל מערך נתונים : 14.07 GiB

  • מבנה תכונה :

FeaturesDict({
    'filename': Text(shape=(), dtype=string),
    'image': Image(shape=(120, 120, 3), dtype=uint8),
    'labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=43)),
    'metadata': FeaturesDict({
        'acquisition_date': Text(shape=(), dtype=string),
        'coordinates': FeaturesDict({
            'lrx': int64,
            'lry': int64,
            'ulx': int64,
            'uly': int64,
        }),
        'projection': Text(shape=(), dtype=string),
        'tile_source': Text(shape=(), dtype=string),
    }),
})
  • תיעוד תכונה :
תכונה מעמד צוּרָה Dtype תיאור
FeaturesDict
שם קובץ טֶקסט חוּט
תמונה תמונה (120, 120, 3) uint8
תוויות Sequence (ClassLabel) (אף אחד,) int64
מטא נתונים FeaturesDict
metadata/acquisition_date טֶקסט חוּט
מטא נתונים/קואורדינטות FeaturesDict
metadata/coordinates/lrx מוֹתֵחַ int64
metadata/coordinates/lry מוֹתֵחַ int64
metadata/coordinates/ulx מוֹתֵחַ int64
metadata/coordinates/uly מוֹתֵחַ int64
מטא נתונים/הקרנה טֶקסט חוּט
metadata/tile_source טֶקסט חוּט

רְאִיָה

bigearthnet/הכל

  • תיאור תצורה : 13 ערוצי Sentinel-2

  • גודל ערכת נתונים: 176.63 GiB

  • מבנה תכונה :

FeaturesDict({
    'B01': Tensor(shape=(20, 20), dtype=float32),
    'B02': Tensor(shape=(120, 120), dtype=float32),
    'B03': Tensor(shape=(120, 120), dtype=float32),
    'B04': Tensor(shape=(120, 120), dtype=float32),
    'B05': Tensor(shape=(60, 60), dtype=float32),
    'B06': Tensor(shape=(60, 60), dtype=float32),
    'B07': Tensor(shape=(60, 60), dtype=float32),
    'B08': Tensor(shape=(120, 120), dtype=float32),
    'B09': Tensor(shape=(20, 20), dtype=float32),
    'B11': Tensor(shape=(60, 60), dtype=float32),
    'B12': Tensor(shape=(60, 60), dtype=float32),
    'B8A': Tensor(shape=(60, 60), dtype=float32),
    'filename': Text(shape=(), dtype=string),
    'labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=43)),
    'metadata': FeaturesDict({
        'acquisition_date': Text(shape=(), dtype=string),
        'coordinates': FeaturesDict({
            'lrx': int64,
            'lry': int64,
            'ulx': int64,
            'uly': int64,
        }),
        'projection': Text(shape=(), dtype=string),
        'tile_source': Text(shape=(), dtype=string),
    }),
})
  • תיעוד תכונה :
תכונה מעמד צוּרָה Dtype תיאור
FeaturesDict
B01 מוֹתֵחַ (20, 20) לצוף32
B02 מוֹתֵחַ (120, 120) לצוף32
B03 מוֹתֵחַ (120, 120) לצוף32
B04 מוֹתֵחַ (120, 120) לצוף32
B05 מוֹתֵחַ (60, 60) לצוף32
B06 מוֹתֵחַ (60, 60) לצוף32
B07 מוֹתֵחַ (60, 60) לצוף32
B08 מוֹתֵחַ (120, 120) לצוף32
B09 מוֹתֵחַ (20, 20) לצוף32
B11 מוֹתֵחַ (60, 60) לצוף32
B12 מוֹתֵחַ (60, 60) לצוף32
B8A מוֹתֵחַ (60, 60) לצוף32
שם קובץ טֶקסט חוּט
תוויות Sequence (ClassLabel) (אף אחד,) int64
מטא נתונים FeaturesDict
metadata/acquisition_date טֶקסט חוּט
מטא נתונים/קואורדינטות FeaturesDict
metadata/coordinates/lrx מוֹתֵחַ int64
metadata/coordinates/lry מוֹתֵחַ int64
metadata/coordinates/ulx מוֹתֵחַ int64
metadata/coordinates/uly מוֹתֵחַ int64
מטא נתונים/הקרנה טֶקסט חוּט
metadata/tile_source טֶקסט חוּט