Datasets:
Formats:
parquet
Size:
10K - 100K
Update README.md
Browse files
README.md
CHANGED
|
@@ -30,7 +30,7 @@ dataset_info:
|
|
| 30 |
dtype: int64
|
| 31 |
- name: x
|
| 32 |
dtype: int64
|
| 33 |
-
- name: y
|
| 34 |
dtype: int64
|
| 35 |
- name: trees
|
| 36 |
list:
|
|
@@ -59,6 +59,19 @@ configs:
|
|
| 59 |
data_files:
|
| 60 |
- split: train
|
| 61 |
path: data/train-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
---
|
| 63 |
|
| 64 |
|
|
@@ -75,11 +88,6 @@ This dataset can be used for various tasks such as tree detection, classificatio
|
|
| 75 |
|
| 76 |
## Dataset Details
|
| 77 |
|
| 78 |
-
### Dataset Description
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
### Metadata Description
|
| 84 |
|
| 85 |
Each JSON file in the `metadata/` directory contains information about the trees in the corresponding satellite image tile.
|
|
@@ -107,6 +115,7 @@ Structure of json object:
|
|
| 107 |
- `radius`: Radius of the tree in pixels, calculated from the area.
|
| 108 |
|
| 109 |
|
|
|
|
| 110 |
|
| 111 |
```json
|
| 112 |
{
|
|
@@ -188,7 +197,7 @@ Of course you can extract more info about trees and place them into the image by
|
|
| 188 |
|
| 189 |
### Dataset Sources
|
| 190 |
|
| 191 |
-
[
|
| 192 |
|
| 193 |
|
| 194 |
## Dataset Creation
|
|
@@ -197,9 +206,13 @@ Dataset was generated by iterating the maximum possible zoom of tile for chosen
|
|
| 197 |
We downloaded each tile as an image with 1024x1024 resolution. We calculated the lat long coordinates for futher calculations.
|
| 198 |
After having corners of tile we could get trees details from public api. We had to make some transformations to be able to draw trees on the images.
|
| 199 |
|
| 200 |
-
|
| 201 |
### Annotation process
|
| 202 |
|
| 203 |
-
We believe that ground truth annotations are legit.
|
| 204 |
[GIS Wroclaw](https://geoportal.wroclaw.pl/mapy/ortofoto/)
|
| 205 |
-
[Trees data](https://mapadrzew.com/)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
dtype: int64
|
| 31 |
- name: x
|
| 32 |
dtype: int64
|
| 33 |
+
- name: 'y'
|
| 34 |
dtype: int64
|
| 35 |
- name: trees
|
| 36 |
list:
|
|
|
|
| 59 |
data_files:
|
| 60 |
- split: train
|
| 61 |
path: data/train-*
|
| 62 |
+
tags:
|
| 63 |
+
- geography
|
| 64 |
+
- trees
|
| 65 |
+
- satellite
|
| 66 |
+
- spatial
|
| 67 |
+
- geospatial
|
| 68 |
+
- city
|
| 69 |
+
- tree
|
| 70 |
+
- image
|
| 71 |
+
- wroclaw
|
| 72 |
+
pretty_name: Satellite Trees of Wroclaw city 2024
|
| 73 |
+
size_categories:
|
| 74 |
+
- 10K<n<100K
|
| 75 |
---
|
| 76 |
|
| 77 |
|
|
|
|
| 88 |
|
| 89 |
## Dataset Details
|
| 90 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
### Metadata Description
|
| 92 |
|
| 93 |
Each JSON file in the `metadata/` directory contains information about the trees in the corresponding satellite image tile.
|
|
|
|
| 115 |
- `radius`: Radius of the tree in pixels, calculated from the area.
|
| 116 |
|
| 117 |
|
| 118 |
+
### Metadata Example
|
| 119 |
|
| 120 |
```json
|
| 121 |
{
|
|
|
|
| 197 |
|
| 198 |
### Dataset Sources
|
| 199 |
|
| 200 |
+
- **Repository:** [Code Repository](https://github.com/Filipstrozik/spatial-data)
|
| 201 |
|
| 202 |
|
| 203 |
## Dataset Creation
|
|
|
|
| 206 |
We downloaded each tile as an image with 1024x1024 resolution. We calculated the lat long coordinates for futher calculations.
|
| 207 |
After having corners of tile we could get trees details from public api. We had to make some transformations to be able to draw trees on the images.
|
| 208 |
|
|
|
|
| 209 |
### Annotation process
|
| 210 |
|
| 211 |
+
We believe that ground truth annotations are legit. (More inside github repository)
|
| 212 |
[GIS Wroclaw](https://geoportal.wroclaw.pl/mapy/ortofoto/)
|
| 213 |
+
[Trees data](https://mapadrzew.com/)
|
| 214 |
+
|
| 215 |
+
# Authors
|
| 216 |
+
- Filip Strózik
|
| 217 |
+
- Dawid Wolkiewicz
|
| 218 |
+
- Izabela Majchrowska
|