metadata
license: unknown
task_categories:
- image-classification
- visual-question-answering
tags:
- numerosity
- counting
- synthetic
- real-world
configs:
- config_name: counting_synthetic
data_files: counting-training/synthetic/*.parquet
- config_name: counting_real_world
data_files: counting-training/real-world/counting-rw-*.parquet
- config_name: ans_synthetic
data_files: ans-training/synthetic/*.parquet
- config_name: ans_real_world
data_files: ans-training/real-world/ans-rw-*.parquet
87cc2s/training
Numerosity training data, organized by task then by source:
counting-training/
synthetic/ <- abstract shapes generator
real-world/ <- real photographs, ~19 public counting/detection datasets
ans-training/
synthetic/ <- abstract shapes generator
real-world/ <- same-category pairs of real photographs
Each *-training/<domain>/ folder has its own dataset_card.md (schema +
generation/curation details), manifest.json (summary stats), and
index.parquet (example_id -> shard/row lookup).