README.md

metadata

dataset_info:
  features:
    - name: images
      sequence: image
    - name: problem
      dtype: string
    - name: answer
      dtype: string
  splits:
    - name: train
      num_bytes: 106400000
      num_examples: 500
  download_size: 106400000
  dataset_size: 106400000
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Stepcount-500

Dataset Description

Stepcount-500 is a benchmark dataset of 500 counting samples with object counts ranging from 11 to 50.

Dataset Structure

Data Fields

• images: A sequence of images (one image per sample)
• problem: The question text describing the counting task
• answer: The ground truth count number

Data Splits

• train: 500 evaluation samples

Dataset Statistics

• Count range: 11-50
• Distribution: 11-20 (130), 21-30 (130), 31-40 (120), 41-50 (120)
• Total dataset size: approximately 106 MB

Usage

from datasets import load_dataset
ds = load_dataset("SI-Lab/Stepcount-500")