HamedAlemo
commited on
Commit
•
8e3da83
1
Parent(s):
8818cb0
update README
Browse files
README.md
CHANGED
@@ -1,9 +1,18 @@
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
4 |
|
5 |
|
6 |
-
# Dataset Card for
|
7 |
|
8 |
## Dataset Description
|
9 |
|
@@ -12,16 +21,16 @@ license: cc-by-4.0
|
|
12 |
|
13 |
### Dataset Summary
|
14 |
|
15 |
-
This dataset contains temporal Harmonized Landsat-Sentinel imagery of diverse land cover and crop type classes across the
|
16 |
|
17 |
-
|
18 |
|
19 |
|
20 |
-
## TIFF
|
21 |
-
Each tiff file
|
22 |
|
23 |
## Band Order
|
24 |
-
|
25 |
Channel, Name, HLS S30 Band number
|
26 |
1, Blue, B02
|
27 |
2, Green, B03
|
@@ -47,9 +56,18 @@ Masks are a single band with values:
|
|
47 |
13 : "Other"
|
48 |
|
49 |
## Class Distribution
|
50 |
-
|
|
|
|
|
|
|
|
|
51 |
|
52 |
## Data Splits
|
53 |
-
The 3,854 chips have been randomly split into training (80%) and validation (20%) and corresponding ids recorded in cvs files `
|
54 |
|
55 |
## Dataset Creation
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc-by-4.0
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
tags:
|
6 |
+
- remote sensing
|
7 |
+
- segmentation
|
8 |
+
- crop type
|
9 |
+
- foundation model
|
10 |
+
size_categories:
|
11 |
+
- 1K<n<10K
|
12 |
---
|
13 |
|
14 |
|
15 |
+
# Dataset Card for Multi-Temporal Crop Classification
|
16 |
|
17 |
## Dataset Description
|
18 |
|
|
|
21 |
|
22 |
### Dataset Summary
|
23 |
|
24 |
+
This dataset contains temporal Harmonized Landsat-Sentinel imagery of diverse land cover and crop type classes across the Contiguous United States for the year 2022. The target labels are derived from USDA's Crop Data Layer (CDL). It's primary purpose is for training segmentation geospatial machine learning models.
|
25 |
|
26 |
+
### Dataset Structure
|
27 |
|
28 |
|
29 |
+
## TIFF Files
|
30 |
+
Each tiff file covers a 224 x 224 pixel area at 30m spatial resolution. Each input satellite file contains 18 bands including 6 spectral bands for three time steps stacked together. Each GeoTIFF file for the mask contains one band with the target classes for each pixel.
|
31 |
|
32 |
## Band Order
|
33 |
+
In each input GeoTIFF the following bands are repeated three times for three observations throughout the growing season:
|
34 |
Channel, Name, HLS S30 Band number
|
35 |
1, Blue, B02
|
36 |
2, Green, B03
|
|
|
56 |
13 : "Other"
|
57 |
|
58 |
## Class Distribution
|
59 |
+
### Training Data Distribution
|
60 |
+
![Training Data](training_dst.png)
|
61 |
+
|
62 |
+
### Validation Data Distribution
|
63 |
+
![Validation Data](validation_dst.png)
|
64 |
|
65 |
## Data Splits
|
66 |
+
The 3,854 chips have been randomly split into training (80%) and validation (20%) and corresponding ids recorded in cvs files `train_data.txt` and `validation_data.txt`.
|
67 |
|
68 |
## Dataset Creation
|
69 |
+
### Query and Scene Selection
|
70 |
+
First, a set of 5,000 chips are defined based on samples from the USDA CDL to ensure a representative sampling across the CONUS. Next, for each chip the corresponding HLS S30 scenes between March and September 2022 are queried and scenes with low cloud cover are retrieved. Then, three scenes are selected among the low cloudy scenes to ensure we get a scene early in the season, one in the middle, and one toward the end. The three final scenes are then reprojected to CDL's projection grid (`EPSG:5070`) using bilinear interpolation.
|
71 |
+
|
72 |
+
### Chip Generation
|
73 |
+
In the final step, the three scenes for each chip are clipped to the bounding box of the chip, and 18 spectral bands are stacked together. In addition, a quality control is applied to each chip using the `Fmask` layer of HLS dataset. Any chip containing clouds, cloud shadow, adjacent to cloud or missing values is discarded. This resulted in 3,854 chips.
|