Datasets:
Package dataset with card and manifest
Browse files- README.md +62 -0
- yolo_birds_1600x896.tar.gz → data/yolo_birds_1600x896.tar.gz +0 -0
- manifest.json +21 -0
README.md
ADDED
|
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pretty_name: Birds Object Detection 1600x896
|
| 3 |
+
task_categories:
|
| 4 |
+
- object-detection
|
| 5 |
+
size_categories:
|
| 6 |
+
- 10K<n<100K
|
| 7 |
+
tags:
|
| 8 |
+
- birds
|
| 9 |
+
- yolo
|
| 10 |
+
- ultralytics
|
| 11 |
+
- object-detection
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# Birds Object Detection 1600x896
|
| 15 |
+
|
| 16 |
+
A single-class bird object-detection dataset prepared for Ultralytics at a
|
| 17 |
+
1600 x 896 camera resolution. The dataset uses a deterministic 90% training
|
| 18 |
+
and 10% validation split.
|
| 19 |
+
|
| 20 |
+
## Contents
|
| 21 |
+
|
| 22 |
+
- `data/yolo_birds_1600x896.tar.gz`: complete Ultralytics dataset archive.
|
| 23 |
+
- `manifest.json`: archive size, SHA-256 checksum, and dataset metadata.
|
| 24 |
+
|
| 25 |
+
After extraction, the archive contains `data.yaml` and matching image and YOLO
|
| 26 |
+
label trees:
|
| 27 |
+
|
| 28 |
+
```text
|
| 29 |
+
data.yaml
|
| 30 |
+
images/train/
|
| 31 |
+
images/val/
|
| 32 |
+
labels/train/
|
| 33 |
+
labels/val/
|
| 34 |
+
```
|
| 35 |
+
|
| 36 |
+
## Download
|
| 37 |
+
|
| 38 |
+
```python
|
| 39 |
+
from huggingface_hub import hf_hub_download
|
| 40 |
+
|
| 41 |
+
archive = hf_hub_download(
|
| 42 |
+
repo_id="PBatch23888/birds-object-detection-1600x896",
|
| 43 |
+
repo_type="dataset",
|
| 44 |
+
filename="data/yolo_birds_1600x896.tar.gz",
|
| 45 |
+
)
|
| 46 |
+
```
|
| 47 |
+
|
| 48 |
+
Private repositories require an authenticated Hugging Face account with
|
| 49 |
+
access to the dataset.
|
| 50 |
+
|
| 51 |
+
## Dataset composition
|
| 52 |
+
|
| 53 |
+
The data builder combines bird annotations from Open Images, COCO 2017,
|
| 54 |
+
Pascal VOC 2012, Birdsnap, and NABirds. Open Images group-of and depiction
|
| 55 |
+
annotations are excluded. All retained annotations are mapped to the single
|
| 56 |
+
class `bird`.
|
| 57 |
+
|
| 58 |
+
## Licensing
|
| 59 |
+
|
| 60 |
+
This is a compilation of multiple upstream datasets. Their respective terms
|
| 61 |
+
and licenses continue to apply. Review each upstream dataset's license before
|
| 62 |
+
redistributing or using this compilation.
|
yolo_birds_1600x896.tar.gz → data/yolo_birds_1600x896.tar.gz
RENAMED
|
File without changes
|
manifest.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"format_version": 1,
|
| 3 |
+
"format": "ultralytics-yolo",
|
| 4 |
+
"class_names": [
|
| 5 |
+
"bird"
|
| 6 |
+
],
|
| 7 |
+
"image_size": [
|
| 8 |
+
896,
|
| 9 |
+
1600
|
| 10 |
+
],
|
| 11 |
+
"split": {
|
| 12 |
+
"train": 0.9,
|
| 13 |
+
"validation": 0.1
|
| 14 |
+
},
|
| 15 |
+
"image_count": 80769,
|
| 16 |
+
"archive": {
|
| 17 |
+
"path": "data/yolo_birds_1600x896.tar.gz",
|
| 18 |
+
"size": 17198845261,
|
| 19 |
+
"sha256": "a3c63269914fd968fdf5f22b75a886a9e2e4fb66b24857dc8c991e6236bf50c3"
|
| 20 |
+
}
|
| 21 |
+
}
|