|
|
--- |
|
|
dataset_info: |
|
|
features: |
|
|
- name: image |
|
|
dtype: image |
|
|
- name: keypoints |
|
|
sequence: |
|
|
sequence: |
|
|
sequence: float64 |
|
|
- name: kp_normalized |
|
|
sequence: |
|
|
sequence: |
|
|
sequence: float64 |
|
|
- name: ground_truth |
|
|
sequence: string |
|
|
splits: |
|
|
- name: train |
|
|
num_bytes: 14998666785.168 |
|
|
num_examples: 23908 |
|
|
- name: test |
|
|
num_bytes: 251316261.464 |
|
|
num_examples: 1056 |
|
|
download_size: 14469845358 |
|
|
dataset_size: 15249983046.632 |
|
|
configs: |
|
|
- config_name: default |
|
|
data_files: |
|
|
- split: train |
|
|
path: data/train-* |
|
|
- split: test |
|
|
path: data/test-* |
|
|
--- |
|
|
|
|
|
A small, referee pose dataset recorded in our lab and at RoboCup German Open 2025. |
|
|
|
|
|
Each sample contains COCO-format pose keypoints with a gesture label; the included image is provided only as a visual reference for the pose. |
