metadata
dataset_info:
features:
- name: image_id
dtype: int64
- name: image
dtype: image
- name: objects
struct:
- name: area
sequence: int64
- name: bbox
sequence:
sequence: int64
- name: id
sequence: int64
- name: iscrowd
sequence: int64
- name: label
sequence: float64
- name: paligemma_labels
dtype: string
splits:
- name: train
num_bytes: 21779064
num_examples: 614
- name: validation
num_bytes: 9092691
num_examples: 264
- name: test
num_bytes: 7562817
num_examples: 223
download_size: 38241726
dataset_size: 38434572
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*