Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
The dataset viewer is not available for this split.
Server error while post-processing the rows. Please report the issue.
Error code:   RowsPostProcessingError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Ancient-Bench: A Comprehensive Multi-millennial, Multi-medium, and Multi-script Benchmark for Ancient Chinese Artifact Text Recognition

Paper GitHub Repository Baidu Netdisk


๐Ÿšจ Important Notice / ๆ ธๅฟƒๅฃฐๆ˜Ž

[EN] Strictly for Academic Research Use Only.
The images in Ancient-Bench are curated from publicly accessible archives of official museums, libraries, and cultural heritage institutions. The dataset creators claim no copyright over the source images. Any commercial exploitation is strictly prohibited. Users must comply with the copyright guidelines of the originating institutions.

[ZH] ๆœฌๆ•ฐๆฎ้›†ไป…้™้žๅ•†ไธšๅญฆๆœฏ็ ”็ฉถไฝฟ็”จใ€‚
Ancient-Bench ๆ•ฐๆฎ้›†ๆ‰€ๆ”ถๅฝ•ๅ›พๅƒๅ‡ๆบ่‡ชๅ„ๅคงๅ›ฝๅฎถ็บงๅš็‰ฉ้ฆ†ใ€ๅ›พไนฆ้ฆ†็ญ‰ๅ…ฌๅผ€ๅ‘ๅธƒ็š„ๆ•ฐๅญ—ๅŒ–ๆ–‡ๅš่ต„ๆบใ€‚ไฝœ่€…ๅ›ข้˜Ÿไธๆ‹ฅๆœ‰ไบฆไธ่ฝฌ่ฎฉไปปไฝ•ๅŽŸๅง‹ๅ›พๅƒ็‰ˆๆƒใ€‚ไธฅ็ฆ็”จไบŽไปปไฝ•ๅ•†ไธš่ฅๅˆฉ่กŒไธบใ€‚


๐Ÿ“Œ Dataset Summary

Ancient-Bench (Accepted by EMNLP 2026) is the first comprehensive benchmark for ancient Chinese artifact text recognition, featuring:

  • Multi-millennial: Spans over 3,000 years (1200 BCE โ€“ 1911 CE) across 3 distinct historical epochs.
  • Multi-medium: Encompasses 9 artifact types (Oracle Bones, Bronzes, Bamboo/Wooden Slips, Silk Manuscripts, Seals, Steles, Cliff Inscriptions, Ancient Editions, and Calligraphy).
  • Multi-script: Covers all 7 canonical historical script styles (Oracle Bone, Bronze, Seal, Clerical, Regular, Cursive, and Running script).
  • Standardized Protocols: Establishes unified standards for medium-specific symbols, paleographic transcriptions (้šถๅฎš), variant characters, and hierarchical layout parsing.

Ancient-Bench Overview


๐Ÿ“Š Dataset Statistics

The benchmark contains 2,700 meticulously annotated high-resolution images from 14 national-level cultural institutions:

Ancient-Bench Dataset Statistics

Medium Category Number of Images Resolution Range (px) Chars / Image Covered Scripts
Oracle (็”ฒ้ชจ) 300 600ร—800 1โ€“68 Oracle Bone Script
Bronze (้’้“œๅ™จ/้‡‘ๆ–‡) 250 41ร—80 โ€“ 1024ร—1024 1โ€“77 Bronze Script
Slip (็ฎ€็‰) 300 29ร—224 โ€“ 700ร—6820 1โ€“107 Seal, Clerical Script
Silk (ๅธ›ไนฆ) 300 46ร—124 โ€“ 1330ร—2067 1โ€“784 Seal, Clerical Script
Seal (ๅฐ็ซ ) 350 100ร—150 โ€“ 1333ร—1341 1โ€“40 Oracle, Bronze, Seal Script
Steles (็ข‘ๅธ–ๅข“ๅฟ—) 300 85ร—155 โ€“ 5795ร—16745 4โ€“816 Seal, Clerical, Regular, Running
Cliff (ๆ‘ฉๅด–็Ÿณๅˆป) 200 183ร—113 โ€“ 7300ร—5760 2โ€“216 Seal, Clerical, Regular, Running
Edition (ๅค็ฑๅˆปๆœฌ) 300 751ร—506 โ€“ 970ร—2817 4โ€“776 Seal, Clerical, Regular, Cursive, Running
Calligraphy (ไนฆๆณ•ๆ‰‹่ฟน) 400 303ร—509 โ€“ 4802ร—13385 2โ€“1666 Bronze, Seal, Clerical, Regular, Cursive, Running
Total 2,700 41ร—80 โ€“ 5795ร—16745 1โ€“1666 All 7 Scripts

๐Ÿš€ How to Download & Access

1. Direct Download via Python (Hugging Face)

You can download the dataset archive using the huggingface_hub Python library:

from huggingface_hub import hf_hub_download

zip_path = hf_hub_download(
    repo_id="hiuyi/Ancient-Bench",
    filename="Ancient-Bench.zip",
    repo_type="dataset",
    local_dir="./",
)

print(f"Dataset downloaded to: {zip_path}")
Downloads last month
54

Paper for hiuyi/Ancient-Bench