Table 1 List of existing Land-Use and Land-Cover (LULC) datasets ready for training Deep Learning (DL) models.

From: Sentinel2GlobalLULC: A Sentinel-2 RGB image tile dataset for global land use/cover mapping with deep learning

Dataset

Source

Source mapping type

Number of images

Image Size

Spatial Resolution

No. Bands

No. Classes

Extent

ISPRS Vaihingen56

Airborne

33 im

2000 × 2000

0.09

3

6

Local

ISPRS Postdam56

Airborne

38 im

6000 × 6000

0.09

3

6

Local

Brazilian coffee scenes57

SPOT-5

Spaceborne

50,004 im

64 × 64

10

3

3

Local

SAT-458

NAIP program

Airborne

500,000 im

28 × 28

1

4

4

Local

SAT-658

NAIP program

Airborne

405,000 im

28 × 28

1

4

6

Local

UCMerced59

OPLS

Airborne

2100 im

256 × 256

0.3

4

21

Local

Zeebruges (link)

LiDAR

Airborne

100,000 im

10 × 10

0.05

3

8

Local

WHU-RS1960

Google Earth

Airborne

1005 im

600 × 600

Up to 0.5

3

19

Local

SIRI-WHU61

Google Earth

Airborne

2.240 im

200 × 200

2

3

12

Local

RSSCN762

Google Earth

Airborne

2800 im

400 × 400

3

7

Local

RSC11 (link)

Google Earth

Airborne

1232 im

512 × 512

0.2

3

11

Local

NWPU-RESISC4518

31,500 im

256 × 256

\(\widetilde{3}0\)–0.2

3

45

Local

AID63

Google Earth

Airborne

10,000 im

600 × 600

\(\widetilde{8}\)-0.5

3

30

Local

BigEarthNet19

Sentinel-2

Satellite

590,326 img.

10 European countries

SpaceNet-764

Dove Satellite Constellation Planet Labs’

Satellite

img.

100 cities