maryzhang/vessels-dataset

Chinese Porcelain Vessels Dataset

Dataset Description

Dataset Summary

The Chinese Porcelain Vessels Dataset is a curated collection of 564 high-resolution images of Chinese porcelain vessels, compiled from manually collected data, public-domain museum archives, and open databases. This foundational dataset is designed for research and educational purposes, supporting motif detection, segmentation, visual analysis, and the creation of derivative datasets for studying traditional Chinese decorative arts.

Supported Tasks and Leaderboards

• Object Detection: Detecting and localizing porcelain vessels in images
• Pattern Recognition: Identifying decorative motifs and patterns
• Image Segmentation: Separating vessels from backgrounds
• Style Classification: Categorizing vessels by dynasty, region, or decorative style
• Cultural Heritage Analysis: Supporting research in art history and cultural preservation
• Dataset Creation: Serving as source material for specialized derivative datasets

Languages

The dataset primarily contains visual data with metadata in English. Pattern descriptions and cultural context may reference Chinese artistic terminology.

Dataset Structure

Data Instances

Each instance in the dataset consists of:

• A high-resolution image of a Chinese porcelain vessel
• Associated metadata (when available) including source attribution

Example structure:

{
  "image": <PIL.Image>,
  "source": "smithsonian_open_access",
  "vessel_type": "vase",
  "file_name": "vessel_001.jpg"
}

Data Fields

• image: PIL Image object containing the vessel photograph
• source: String indicating the data source (smithsonian, met, npm, freepik, manual_collection)
• vessel_type: String indicating vessel category (when labeled)
• file_name: Original filename

Data Splits

The dataset is provided as a single training split with 564 images. Users are encouraged to create their own train/validation/test splits based on their specific use case.

Dataset Creation

Curation Rationale

This dataset was created to address the need for high-quality, openly accessible images of Chinese porcelain for computer vision research, digital humanities projects, and educational applications. The focus on Chinese porcelain vessels provides a rich domain for studying:

• Traditional decorative patterns and motifs
• Cultural heritage preservation through digital methods
• Cross-cultural artistic influence and trade history
• Evolution of ceramic techniques and styles

Source Data

Initial Data Collection and Normalization

Images were collected from multiple sources to ensure diversity in:

• Historical periods (various Chinese dynasties)
• Vessel types (vases, bowls, plates, cups, bottles)
• Decorative styles (blue and white, famille rose, celadon, etc.)
• Photographic conditions (museum lighting, various angles)

Resolution and format normalization was applied while preserving original image quality.

Who are the source language producers?

The vessels themselves were created by Chinese artisans across various dynasties. The photographs were produced by:

• Museum photographers and digitization specialists
• Cultural heritage documentation teams
• Open-source contributors

Annotations

Annotation process

The dataset currently provides minimal annotations, focusing on source attribution and basic categorization. Users can add their own annotations for specific tasks.

Who are the annotators?

Initial categorization was performed by the dataset creator with reference to museum cataloging information where available.

Personal and Sensitive Information

The dataset contains no personal or sensitive information. All images are either in the public domain or used under appropriate open access licenses.

Considerations for Using the Data

Social Impact of Dataset

This dataset contributes to:

• Cultural Preservation: Digital preservation of Chinese artistic heritage
• Educational Access: Democratizing access to museum-quality cultural artifacts
• Research Advancement: Enabling computational approaches to art history
• Cross-Cultural Understanding: Facilitating appreciation of Chinese decorative arts

Discussion of Biases

Potential biases in the dataset include:

• Temporal Bias: Certain dynasties may be overrepresented based on museum collection priorities
• Style Bias: Blue and white porcelain may be overrepresented due to its popularity in Western collections
• Geographic Bias: Sources primarily from Western museums may not fully represent pieces held in Asian institutions
• Preservation Bias: Better-preserved or more valuable pieces are more likely to be photographed and digitized
• Photographic Bias: Museum photography standards may not represent how these objects appear in natural settings

Other Known Limitations

• Limited metadata about specific dynasties, kilns, or regional variations
• Varying image quality and lighting conditions across sources
• Some images may include museum display elements (stands, cases)
• Not all vessel types are equally represented
• Modern reproductions may be present without clear labeling

Additional Information

Dataset Curators

Mary Zhang

Licensing Information

Dataset License

This dataset compilation is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Source Licenses

Individual images are subject to their source licenses:

• Smithsonian Open Access: CC0 (Public Domain)
• Metropolitan Museum of Art: CC0 (Public Domain)
• National Palace Museum: Check individual image rights
• Freepik: Freepik License (attribution required)
• Manual Collection: Various (documented per image where applicable)

Users must comply with the specific license requirements of each source when using images from this dataset.

Citation Information

BibTeX:

@dataset{zhang2024vessels,
  author = {Zhang, Mary},
  title = {Chinese Porcelain Vessels Dataset},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co/datasets/maryzhang/vessels-dataset},
  note = {A collection of 564 high-resolution images of Chinese porcelain vessels from museum archives and open databases}
}

APA: Zhang, M. (2024). Chinese Porcelain Vessels Dataset [Dataset]. HuggingFace. https://huggingface.co/datasets/maryzhang/vessels-dataset

Contributions

Thanks to the following institutions for making their collections openly accessible:

• Smithsonian Institution for their Open Access initiative
• The Metropolitan Museum of Art for The Met Open Access program
• National Palace Museum for their Digital Archive
• Freepik for vector and pattern resources
• All museums and institutions working to digitize and share cultural heritage

Dataset Sources

Primary Sources

1. Smithsonian Institution

   • Collection: Smithsonian Open Access Collections
   • URL: https://www.si.edu/openaccess
   • License: CC0 (Public Domain)
   • Content: High-resolution photographs of Asian art collections

2. The Metropolitan Museum of Art

   • Collection: The Met Open Access Collection
   • URL: https://www.metmuseum.org/art/collection
   • License: CC0 (Public Domain)
   • Content: Comprehensive ceramic and porcelain collections

3. National Palace Museum Digital Archive

   • Collection: Porcelain and Ceramics Collection
   • URL: https://theme.npm.edu.tw
   • License: Varies by image
   • Content: Significant Chinese imperial porcelain collection

4. Freepik

   • Collection: Vector and Decorative Pattern Assets
   • URL: https://www.freepik.com
   • License: Freepik License (attribution required)
   • Content: Supplementary decorative patterns and motifs

Usage Guidelines

Recommended Use Cases

• Academic research in art history and digital humanities
• Training computer vision models for cultural heritage applications
• Educational materials for teaching about Chinese decorative arts
• Generating derivative datasets for specific pattern analysis tasks
• Non-commercial creative projects exploring traditional motifs

Not Recommended For

• Commercial reproduction of specific museum pieces without proper licensing
• Authentication or valuation of actual porcelain pieces
• Making claims about provenance or historical attribution without expert verification

Ethical Considerations

This dataset is created with respect for Chinese cultural heritage and the institutions preserving these artifacts. Users should:

• Acknowledge the cultural significance of these objects
• Credit source institutions appropriately
• Consider the colonial history of some museum collections
• Use the dataset to promote cultural understanding and preservation

AI-Assisted Development

• Transparency Note: ChatGPT was utilized during the development process for: code generation and debugging

---