MiikeMineStamps

This is an official webpage of MiikeMineStamps dataset.

The MiikeMineStamps dataset is a naturally long-tailed open-ended dataset in the domain of Japanese historical documents. It contains 5056 images of Japanese stamps that belong to 415 classes including two special ones. The stamps were extracted from a large compendium of historical documents from the Japanese company Mitsui Mi’ike Mine, one of the largest business archives in modern Japan that spans half a century, includes tens of thousands of documents, and has been widely used by labor historians, business historians, and others.

Download Dataset

Info

The current version is v1.0.

The collection currently has 5056 images of 415 stamps collected from hundreds of documents. New versions will be published as our expert labels more images.

Figure 1 shows several examples of stamps.

Example images — Figure 1. Examples of stamps in the dataset.

The histogram of number of samples per stamp name is shown below. The dataset is collected via active learning, colors signify the cycle.

Histogram of stamp names. — Figure 2. Histogram of stamps by name. Note the log scale. Classes with 1 instance are not shown.

License

The dataset is available under Creative Commons V4 license. If you find this work useful, please cite:

@inproceedings{MiikeMineStamps,
  author = {Toropov, Evgeny and Buitrago, Paola and Prabha, Rajanie and Uran, Julian and Adal, Raja},
  year = {2021},
  title = {MiikeMineStamps: A Long-Tailed Dataset of JapaneseStamps via Active Learning},
  journal = {ICDAR '21: 16th International Conference on Document Analysis and Recognition},
}

Dataset structure

The dataset consists of 5056 cropped images of stamps (Figure 1) and the related meta-information. Meta-information is available in two formats: json and SQLite database with Shuffler schema. Json is human-readable and easier to load with conventional tools, Shuffler database format is easier to get various statistics.

The following information is available for each image:

imagefile: the path to the image.
name: stamp class name, see the conventions below.
decade: the decade of the document where the stamp was originally placed.
x_on_page: x coordinate of the stamp center relative to the enclosing page (in range [0, 1]).
y_on_page: y coordinate of the stamp center relative to the enclosing page (in range [0, 1]).
width_on_page: width of the stamp relative to the enclosing page (in range [0, 1]).
height_on_page: height of the stamp relative to the enclosing page (in range [0, 1]).

The last four fields are recorded when pages can be identified in a photograph and the stamp is located inside one of the pages. Some stamp classes are consistently located in one of the corners of their pages.

Stamp names conventions

Stamp names are usually based on the phonetic reading of the Japanese characters using a modified Hepburn system. There are several exceptions to this, however. First, stamps that ask different catefories of users such as branch head, secretary, etc. to certify a document by placing their stamp in a particular place are collectively labeled "stamphere." Second,

Stamp names are lowercase letters, underscores, and two special characters: "+" and "?". Stamp names are usually based on the phonetic reading of the Japanese characters using a modified Hepburn system.
Most names contain only letters, e.g. "kuru".
"+" in a name means that only one half of a stamp is visible, because two conforming documents were stamped together. Only the top of a stamp is visible if "+" is after the stamp name (e.g. 'kei+'), and only the bottom of a stamp is visible if "+" before the stamp name (e.g. '+kei').
There are pairs of identical stamp names, one of which ends with the suffix "_east", e.g. "itou" and "itou_east". These are the names of two different individuals who have the same phonetic reading in English, such as 伊藤 and 伊東, both of which are read as itou. They are referred to as "itou" and "itou_east" in reference to the second Chinese character of the name.
"?" in a name (e.g."oomuta?unhontenin") means that one symbol in a stamp was not recognized.
Name "??" is a special name that means that a stamp was not recognized.
Name "+??" is a special name that means that a stamp was not recognized and only its bottom is visible.
Name "stamphere" is a special name. Stamps that ask different catefories of users such as branch head, secretary, etc. to certify a document by placing their stamp in a particular place are collectively labeled "stamphere."

Code

The code used to collect the dataset is published at https://github.com/pscedu/ml4docs

Funding

NSF grant #1548562 XSEDE 2.0: Integrating, Enabling and Enhancing National Cyberinfrastructure with Expanding Community Involvement
NSF grant #1445606 Bridges: From Communities and Data to Workflows and Insight
NSF grant #1928147 Category I: Bridges-2: Scalable Converged Computing, Data, and Analytics for Rapidly Evolving Science and Engineering Research

Acknowledgements

This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1548562. Specifically, it used the Bridges and Bridges-2 systems, which are supported by NSF award number ACI-1445606 and ACI-1928147, at the Pittsburgh Supercomputing Center (PSC). The work was made possible through the XSEDE Extended Collaborative Support Service (ECSS) program.