Datasets:
Modalities:
Image
Formats:
imagefolder
Size:
1K - 10K
ArXiv:
Tags:
image
calligraphy
chinese-calligraphy
historical-documents
fine-grained-recognition
cultural-heritage
License:
| language: | |
| - zh | |
| - en | |
| license: cc-by-nc-4.0 | |
| size_categories: | |
| - 10K<n<100K | |
| task_categories: | |
| - image-classification | |
| - image-to-text | |
| pretty_name: 'HCSU: Fine-Grained Historical Calligraphy Style Understanding Dataset' | |
| tags: | |
| - image | |
| - calligraphy | |
| - chinese-calligraphy | |
| - historical-documents | |
| - fine-grained-recognition | |
| - cultural-heritage | |
| - style-understanding | |
| # HCSU: A Dataset and Benchmark for Fine-Grained Historical Calligraphy Style Understanding (V1.0) | |
| <div align="center"> | |
| [](https://huggingface.co/papers/2607.04147) | |
| [](https://github.com/209-Tongji/HCSU) | |
| [](https://huggingface.co/datasets/Tongji209/HCSU) | |
| [](https://www.modelscope.cn/datasets/Tongji209/HCSU_Fine-Grained_Historical_Calligraphy_Style_Understanding_Dataset) | |
| [](https://www.apache.org/licenses/LICENSE-2.0) | |
| [](https://creativecommons.org/licenses/by-nc/4.0/) | |
| <p><strong>ECCV 2026</strong></p> | |
| <p>Yinsheng Yao<sup>*</sup>, Yan Liu<sup>*</sup>, Chen Ye<sup>†</sup></p> | |
| <p><sup>*</sup> Equal contribution. <sup>†</sup> Corresponding author.</p> | |
| </div> | |
| > **Abstract:** Addressing the "Knowledgeable but unperceptive" dilemma where existing Large Vision-Language Models (LVLMs) possess historical knowledge but lack fine-grained calligraphy style perception, we introduce **HCSU**—the first large-scale dataset and evaluation benchmark specifically tailored for fine-grained historical calligraphy style understanding. HCSU contains 39,307 meticulously annotated high-definition Chinese character images, accompanied by expert-level hierarchical aesthetic descriptions. Through a pioneering rigorous data processing pipeline, we successfully decouple authentic ink manuscripts (**Tie**) from stone rubbings (**Bei**), thoroughly resolving the "modality aliasing" problem that has long plagued the digital cultural heritage field. | |
| Code is available at: [209-Tongji/HCSU](https://github.com/209-Tongji/HCSU). | |
| <div align="center"> | |
| <img src="./assets/head.png" alt="Figure 1: HCSU Dataset Overview and Multi-dimensional Annotation" width="80%"> | |
| <p><i>Figure 1: The panoramic data experience of HCSU. Beyond single images, we provide technical dimensions of calligraphy expression (Ink Style, Stroke Style, Structure) and expert-level descriptions of aesthetic spirit, enabling highly interpretable fine-grained visual reasoning.</i></p> | |
| </div> | |
| ## 📊 1. Dataset Overview | |
| Traditional OCR or Document AI datasets typically focus only on "content recognition." HCSU marks a paradigm shift by focusing the core task on **fine-grained perception of artistic style**. To eliminate interference caused by different physical media in historical archives, HCSU implements strict domain partitioning during its construction. | |
| ### Core Dataset Statistics | |
| - **Total Scale:** 39,307 high-resolution character images (normalized to 256×256) subjected to rigorous manual verification. | |
| - **Diversity Coverage:** Spans 10 major historical dynasties (from the Three Kingdoms to the Modern era), covering 49 calligraphic masters who serve as milestones in Chinese art history. | |
| - **Script Classification:** Full coverage of the 5 major scripts: Seal (Zhuanshu), Clerical (Lishu), Regular (Kaishu), Semi-cursive (Xingshu), and Cursive (Caoshu). | |
| - **Three Sub-domains:** | |
| - 🖌️ **Ink Manuscripts (Tie):** 3,780 images, highly preserving authentic ink dynamics (e.g., ink wash, "flying white" textures). | |
| - 🪨 **Stone Rubbings (Bei):** 3,240 images, extracting pure geometric structures and stroke tension. | |
| - 📜 **Wild Data:** 32,287 images, raw character images from calligraphy works without the following data processing pipeline. | |
| - **Archive Passwords:** | |
| - `bei.zip`: `eWZNA9hzJoyMHtQwVh7QnBtN7` | |
| - `tie.zip`: `eWZNA9hzJoyMHtQwVh7QnBtN7` | |
| - `wild.zip.001` to `wild.zip.006`: To request the password, email `yechen@tongji.edu.cn` with your institution, professional title, and a brief description of the intended use. | |
| - **Annotations:** Full metadata for the released Bei/Tie archives is provided in `bei_annotations.json` and `tie_annotations.json`. | |
| - **Example Subsets:** `bei_samples/` and `tie_samples/` are provided uncompressed for quick preview. | |
| ### Repository File Tree | |
| ```text | |
| . | |
| ├── README.md | |
| ├── bei.zip | |
| ├── tie.zip | |
| ├── wild.zip.001 | |
| ├── wild.zip.002 | |
| ├── wild.zip.003 | |
| ├── wild.zip.004 | |
| ├── wild.zip.005 | |
| ├── wild.zip.006 | |
| ├── bei_annotations.json | |
| ├── tie_annotations.json | |
| ├── bei_samples/ | |
| ├── tie_samples/ | |
| └── assets/ | |
| ``` | |
| <div align="center"> | |
| <img src="./assets/data.png" alt="Figure 2: Comprehensive Statistics of the HCSU Dataset" width="100%"> | |
| <p><i>Figure 2: Comprehensive statistics of the HCSU dataset. Displays the distribution of script types across the three domains (Wild, Bei, Tie) and a stacked chart of character counts by dynasty.</i></p> | |
| </div> | |
| --- | |
| ## 👁️ 2. Data Experience: Beyond "Flattened Labels" | |
| Existing calligraphy datasets (such as MCCD, CalliNet) usually provide only "flattened labels" like "Author" or "Character ID," which cannot support deep visual interpretation by large models. HCSU introduces a **multi-dimensional, interpretable hierarchical annotation framework** that bridges low-level visual features with high-level aesthetic judgments. | |
| Each instance in HCSU provides the following rich metadata experience: | |
| 1. **Semantic and Contextual Ground Truth:** Character ID (`char`), Calligrapher (`author`), Dynasty (`dynasty`), Source Medium (`source_type`), and preservation `quality`. | |
| 2. **Fine-grained Visual Attribute Decomposition:** | |
| - `ink_style`: Describes ink density, moisture, permeation, and "flying white" (Feibai) textures. | |
| - `stroke_style`: Micro-level brushwork features (e.g., "center-tip brushwork," "hidden tip," "angular turns"). | |
| - `character_structure`: Overall spatial layout and geometric proportions (e.g., "tight center and loose exterior," "dynamic balance between tilted and upright"). | |
| 3. **Expert Natural Language Description:** Professional aesthetic commentary (under 50 words) written by calligraphy connoisseurs, serving as the Ground Truth for generation tasks. | |
| <div align="center"> | |
| <img src="./assets/overview.png" alt="Figure 3: HCSU Dataset Examples" width="100%"> | |
| <p><i>Figure 3: Decoupling style and content. HCSU requires models to identify consistent artistic signatures across visually distinct instances (e.g., a and c belong to the same calligrapher) while distinguishing instances with identical content but different artistic expressions (e.g., a and b).</i></p> | |
| </div> | |
| --- | |
| ## ⚙️ 3. Strict Data Processing Pipeline | |
| To build an evaluation benchmark with absolute structural consistency, free from complex background interference (e.g., paper aging, mounting borders, stone cracks), we designed a multi-stage rigorous data processing pipeline. This is the core competency that allows HCSU to preserve "authentic ink charm" without loss. | |
| ### Phase 1: Domain-Adaptive Polarity Normalization and Visual Heuristic Correction | |
| - **Metadata-driven Polarity Inversion:** Stone rubbings (Bei) typically appear as white characters on a black background. We perform a global polarity inversion ($255 - I_{gray}$) to unify them into a standard "black on white" configuration, followed by Otsu's method to obtain an initial binary map. | |
| - **Visual Heuristic Error Correction:** Metadata in historical archives is often inaccurate. We check the average pixel intensity of four $N \times N$ corner regions. If the intensity is below a strict threshold $\tau_{corner}$ (indicating a dark background), the heuristic algorithm overrides the metadata and automatically executes bit inversion, ensuring zero-human-intervention polarity accuracy. | |
| ### Phase 2: Geometric Lossless Normalization | |
| Traditional direct resizing leads to severe anisotropic deformation, destroying the critical structural tension of calligraphy. | |
| 1. We first embed the original image into the center of a new $S \times S$ pure white canvas ($S$ is the maximum of the original width and height). | |
| 2. We use the high-quality **Lanczos interpolation algorithm** to downsample the square image to the target $256 \times 256$ resolution. This operation effectively suppresses aliasing effects and perfectly maintains the original aspect ratio and stroke sharpness. | |
| ### Phase 3: Morphological Denoising and "Authentic Ink" Appearance Reconstruction | |
| While binary maps provide clean structural priors, they discard ink textures (like Feibai) crucial for expert appreciation. | |
| 1. We apply 8-connected component analysis (CCA) to the binary mask, filtering out independent noise spots smaller than 50 pixels (e.g., paper grain or stone cracks). | |
| 2. The cleaned semantic mask is re-mapped back onto the geometrically aligned original high-resolution RGB image. | |
| 3. **All background pixels are replaced with pure white (`[255, 255, 255]`)**. This mechanism perfectly strips away complex background interference while **preserving 100% of the authentic ink density, stroke texture, and writing pressure variations**. | |
| <div align="center"> | |
| <img src="./assets/sample.png" alt="Figure 4: Unified Data Construction Process" width="80%"> | |
| <p><i>Figure 4: Demonstration of the HCSU data construction pipeline. Top row: Background stripping and ink reconstruction for Ink Manuscripts (Tie); Bottom row: Polarity inversion and denoising for Stone Rubbings (Bei).</i></p> | |
| </div> | |
| --- | |
| ## 🏆 Evaluation Protocols | |
| Based on the processed high-quality data, HCSU proposes two rigorous evaluation tasks (the test set employs strict Anti-Leakage sampling rules, ensuring that target and reference images never contain the same character): | |
| 1. **Fine-grained Style Discrimination (8-way Candidate Selection):** Tests whether the model can correctly pair styles among 1 correct style and 7 strong distractors, relying solely on micro-level brushwork and structural features. | |
| 2. **Explainable Aesthetic Reasoning (Style Description Generation):** Given a 2-shot prompt, the model is required to generate a calligraphic critique under 50 words using professional terminology. We use a dual-track evaluation framework consisting of `BERTScore` and `LLM-as-a-Judge` (evaluating terminology accuracy and detail richness). | |
| *Evaluation code and Prompt templates can be found in our open-source repository `eva`.* | |
| --- | |
| ## 📝 Citation | |
| If you use this dataset in your research or are inspired by our data processing pipeline, please consider citing our paper: | |
| ```bibtex | |
| @inproceedings{hcsu2026, | |
| title={HCSU: A Dataset and Benchmark for Fine-Grained Historical Calligraphy Style Understanding}, | |
| author={Yao, Yinsheng and Liu, Yan and Ye, Chen}, | |
| booktitle={Proceedings of the European Conference on Computer Vision (ECCV)}, | |
| year={2026} | |
| } | |
| ``` | |
| ## 📄 License | |
| This repository uses separate licenses for code and data: | |
| - **Code, scripts, and repository documentation:** [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0). A repository copy is provided in [LICENSE](./LICENSE). | |
| - **Dataset images, archives, annotations, samples, and dataset-derived visual assets:** [Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)](https://creativecommons.org/licenses/by-nc/4.0/). The legal code is available at [creativecommons.org/licenses/by-nc/4.0/legalcode](https://creativecommons.org/licenses/by-nc/4.0/legalcode), and a repository scope note is provided in [DATA_LICENSE](./DATA_LICENSE). | |
| The dataset is provided for academic and non-commercial research use. Commercial use requires separate permission from the rights holders. |