Does anyone have a copy of the ICDAR2013 Chinese Handwriting Competition Dataset? [R]

2026-05-29 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, quick

Summary

A user is seeking the ICDAR2013 Chinese Handwriting Recognition Competition Dataset, as its official download page at "https://nlpr.ia.ac.cn/databases/handwriting/Download.html" has been consistently down for weeks. Extensive searches across platforms like Kaggle, HuggingFace, Google Drive, Baidu Netdisk, and GitHub, along with advanced Google dorking, yielded no direct copies of the original dataset. However, an update reveals a partial solution: the "shannanyinxiang/PageNet" GitHub repository contains the dataset preprocessed into LMDB format. While this version lacks line and word-level annotations, character-level annotations from the "newbie2niubility/TGC-Diff" repository could potentially be used for reconstruction. This reconstruction would need to account for differences in page boundaries and margins present in the TGC-Diff annotations compared to the cropped PageNet LMDB images.

Key takeaway

For AI scientists or NLP engineers seeking the ICDAR2013 Chinese Handwriting Recognition Dataset, be aware that the official source is offline. If you require this dataset, prioritize checking community-maintained GitHub repositories like "shannanyinxiang/PageNet" for preprocessed versions. You may need to combine data from multiple sources, such as "newbie2niubility/TGC-Diff" for character-level annotations, and carefully manage discrepancies in data formatting, like page boundaries and margins, to reconstruct the full dataset.

Key insights

The ICDAR2013 Chinese Handwriting Dataset is difficult to obtain, but a partial LMDB version exists.

Principles

• Official dataset links can become unavailable.
• Community repositories may offer preprocessed data.
• Data reconstruction requires careful annotation alignment.

Method

To reconstruct full annotations for the ICDAR2013 dataset from the PageNet LMDB, align character-level annotations from TGC-Diff, adjusting for page boundaries and margins.

In practice

• Check GitHub for preprocessed dataset versions.
• Cross-reference multiple repositories for annotations.
• Verify annotation alignment for cropped images.

Topics

• ICDAR2013 Dataset
• Chinese Handwriting Recognition
• Dataset Availability
• LMDB Format
• Data Reconstruction
• Annotation Alignment

Code references

• shannanyinxiang/PageNet
• newbie2niubility/TGC-Diff

Best for: Research Scientist, Machine Learning Engineer, AI Scientist, NLP Engineer

Related on AIssential

• Manga109-v2026: Revisiting Manga109 Annotations for Modern Manga Understanding — Revising foundational datasets with precise annotations is crucial for advancing modern multimodal AI understanding tasks.
• Qwen-Image-2.0 renders ancient Chinese calligraphy and PowerPoint slides with near-perfect text accuracy — Qwen-Image-2.0 unifies image generation and editing with exceptional text rendering, including complex calligraphy, at a compact 7B parameters.
• STiTch: Semantic Transition and Transportation in Collaboration for Training-Free Zero-Shot Composed Image Retrieval — Training-free zero-shot CIR improves by refining LLM-generated captions and using set-to-set alignment for diverse compositions.
• Why Nepali OCR Is Brutally Hard — And How We Finally Solved It — A hybrid deep learning pipeline overcomes complex script challenges to automate document data extraction and form pre-filling.
• KazakhOCR: A Synthetic Benchmark for Evaluating Multimodal Models in Low-Resource Kazakh Script OCR — Current MLLMs struggle with OCR and language identification for low-resource Kazakh scripts, especially Arabic, performing worse than traditional OCR.
• A Chinese Education Broadcast Emotion Corpus — A new Chinese education broadcast emotion corpus aids emotion recognition research.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.