Hugging Face's logo

Datasets:
PBatch23888
/
birds-object-detection-1600x896

Tasks:
Object Detection
Modalities:
Tabular
Text
Formats:
json
Size:
< 1K
Tags:
birds
yolo
ultralytics
object-detection
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
1
birds-object-detection-1600x896 / README.md
PBatch23888's picture
PBatch23888
Package dataset with card and manifest
3b5228d verified
preview code
|
Raw
History Blame Contribute Delete
1.53 kB
metadata 
pretty_name: Birds Object Detection 1600x896
task_categories:
  - object-detection
size_categories:
  - 10K<n<100K
tags:
  - birds
  - yolo
  - ultralytics
  - object-detection

Birds Object Detection 1600x896

A single-class bird object-detection dataset prepared for Ultralytics at a 1600 x 896 camera resolution. The dataset uses a deterministic 90% training and 10% validation split.

Contents

  • data/yolo_birds_1600x896.tar.gz: complete Ultralytics dataset archive.
  • manifest.json: archive size, SHA-256 checksum, and dataset metadata.

After extraction, the archive contains data.yaml and matching image and YOLO label trees:

data.yaml
images/train/
images/val/
labels/train/
labels/val/

Download 

from huggingface_hub import hf_hub_download

archive = hf_hub_download(
    repo_id="PBatch23888/birds-object-detection-1600x896",
    repo_type="dataset",
    filename="data/yolo_birds_1600x896.tar.gz",
)

Private repositories require an authenticated Hugging Face account with access to the dataset.

Dataset composition

The data builder combines bird annotations from Open Images, COCO 2017, Pascal VOC 2012, Birdsnap, and NABirds. Open Images group-of and depiction annotations are excluded. All retained annotations are mapped to the single class bird.

Licensing

This is a compilation of multiple upstream datasets. Their respective terms and licenses continue to apply. Review each upstream dataset's license before redistributing or using this compilation.