Drawings Gemma-Enriched Dataset. Fotothek - Bibliotheca Hertziana

Version 1.0
Liuzzo, Pietro Maria, 2025, "Drawings Gemma-Enriched Dataset. Fotothek - Bibliotheca Hertziana", https://doi.org/10.17617/3.1GN3OL, Edmond, V1
Learn about Data Citation Standards.
Contact Owner Share
Dataset Metrics
7 Downloads
Description

Zeichnungen Dataset - AI-Enhanced Art Historical Descriptions

Zeichnungen Dataset - AI-Enhanced Art Historical Descriptions with Iconography

This dataset contains 224x224 images and relative metadata extracted from the MIDAS XML of the Catalogue of the Photographic Collection of the Bibliotheca Hertziana enriched with AI-generated prose texts and iconographic analysis. The dataset is limited to photographs of objects classified as drawing (Zeichnungen), and has been processed using Google Gemma 2 9B Instruct large language model on the KISSKI HPC cluster of the GWDG. Scripts to process the data on KISSKI have been elaborated with Claude Code in Virtual Studio Code.

Dataset Overview

Source Data:

Output:

  • Enriched metadata: TSV files with AI-generated German and English descriptions
  • Iconographic analysis: Descriptions based on ICONCLASS classification
    • 224x224 images downloaded from IIIF Image Api of the Photographic Collection

Processing Pipeline

1. Data Extraction

Source data was extracted with zeichnungen.xql from MIDAS XML format combined.xml containing structured art historical metadata including:

  • Object titles and descriptions (textobj, textfoto)
  • Artist information (aob30)
  • Location data (aob26, aob28)
  • ICONCLASS codes (a5500) - Standardized iconographic classification
  • Dating and provenance
  • Image references (a8540) The set was limited to 30000 entries.

2. ICONCLASS Cache Preparation

ICONCLASS System:

  • Source: ICONCLASS.org - Multilingual classification system for cultural content
  • GitHub repository: https://github.com/iconclass/data
Images Download

224x224 images downloaded in advance from the IIIF Service based on gemalde.tsv. The script processing for AI Text Enrichment from the metadata checks that the image has been downloaded, so the output data has a 100% certainty of having a matching image. 28.165 images downloaded from 29,999 rows. This is due to known missing digital images. The dataset corresponds to published data and each row contains the licence and accessibility of the single image, date of creation and last update of the catalogue object.

3. AI Text Generation

Model Used:

  • Name: Google Gemma 2 9B Instruct
  • Parameters: 9 billion
  • Quantization: FP16 (no quantization)
  • Context window: 8,192 tokens
  • License: Gemma Terms of Use

Processing Workflow:

  1. Input cleaning: Removal of numeric codes, normalization of Unicode characters, increased CSV field size limit (10 MB)
  2. Paragraph generation: German text from structured metadata
  3. ICONCLASS lookup: Offline cache-based iconographic description retrieval
  4. Iconographic synthesis: AI-generated description from ICONCLASS codes
  5. Translation: German → English
  6. Categories processed:
    • paragraph foto DE/EN - Photograph description
    • paragraph obj DE/EN - Object/artwork description
    • paragraph verwalter DE/EN - Collection/custodian information
    • paragraph standort DE/EN - Location information
    • paragraph iconclass DE/EN - Iconographic content description (NEW)

AI Prompts Used

Paragraph Generation Prompt

Convert the following structured information into a coherent text in German.
The text contains field data that should be transformed into flowing prose while preserving all information.

IMPORTANT:
- Write a MAXIMUM of 2 paragraphs
- Do NOT include any URLs or web links
- Do NOT include reference codes or numerical codes
- Do NOT add any comments or explanations
- Only output the paragraph text itself

Field: {field_name}
Text: {cleaned_text}

German text (maximum 2 paragraphs):

ICONCLASS Paragraph Prompt

Based on the following Iconclass descriptions, write a brief German paragraph describing what the image depicts.

Descriptions: {'; '.join(descriptions)}

IMPORTANT:
- Start with "Das Bild zeigt" or similar phrasing
- Combine all descriptions into a flowing text
- Maximum 1-2 sentences
- Do NOT include iconclass codes or numbers
- Do NOT include reference codes starting with "bh"
- Only output the descriptive German text

German description:

Example ICONCLASS Processing:

Input from data:

a5500: 31 A 23 1 | 31 A 25 11 | 31 B 62 11

ICONCLASS lookup (from cache):

31 A 23 1  → "standing figure"
31 A 25 11 → "arm raised upward"
31 B 62 11 → "looking upwards"

AI-generated output (DE):

Das Bild zeigt eine stehende Figur mit erhobenem Arm, die nach oben blickt.

Translation (EN):

The image shows a standing figure with raised arm, looking upwards.

Translation Prompt

Translate the following German text to English.
Preserve the meaning and style as much as possible.

IMPORTANT:
- Do NOT include any URLs or web links in the translation
- Do NOT include reference codes starting with "bh" followed by numbers
- Do NOT include numerical codes like 08012353
- Do NOT add any comments or explanations
- Only output the translated text itself

German text: {text}

English translation:

KISSKI Cluster Resources

Hardware Configuration

GPU: NVIDIA A100 (80GB VRAM)

  • Architecture: Ampere
  • Tensor Cores: 432
  • FP16 Performance: ~312 TFLOPS
  • Memory Bandwidth: 2 TB/s

Allocation per job:

  • GPUs: 1× A100
  • CPUs: 4 cores
  • RAM: 64 GB
  • Time limit: 6 hours per job

Job Array Configuration

Array setup:

  • Total jobs: 75 (indices 0-74)
  • Chunk size: 400 rows per job
  • Parallel jobs: 10 simultaneous
  • Total rows processed: 30,000 (rows 0-29,999)

Output Structure

data_zeichnungen/
├── enriched_data/
│   ├── zeichnungen_0-399.tsv          # Rows 0-399
│   ├── zeichnungen_400-799.tsv        # Rows 400-799
│   ├── zeichnungen_800-1199.tsv       # Rows 800-1199
│   └── ...
├── images/
│   ├── {image_id_1}.jpg               # IIIF thumbnail (224×224)
│   ├── {image_id_2}.jpg
│   └── ...
└── README.md                           # This file

Output Fields

Each TSV file contains the original metadata plus AI-generated fields:

Original fields: All fields from zeichnungen.tsv including:

  • a8540 - Image ID (BILDDATEI-NR.)
  • textobj - Original object text
  • textfoto - Original photo text
  • a5500 - ICONCLASS codes (primäre Ikonographie)
  • aob26, aob28, aob30 - Relations
  • etc.

Generated fields:

  • paragraph foto DE - German description of photograph
  • paragraph foto EN - English translation
  • paragraph obj DE - German description of object/artwork
  • paragraph obj EN - English translation
  • paragraph verwalter DE - German description of collection
  • paragraph verwalter EN - English translation
  • paragraph standort DE - German description of location
  • paragraph standort EN - English translation
  • paragraph iconclass DE - German iconographic description
  • paragraph iconclass EN - English iconographic description

KISSKI Documentation

  • Main documentation: https://docs.hpc.gwdg.de/
  • GPU partitions: https://docs.hpc.gwdg.de/how_to_use/compute_partitions/gpu_partitions/
  • Account types: https://docs.hpc.gwdg.de/start_here/account_types/

ICONCLASS Resources

  • Official website: https://iconclass.org/
  • Data repository: https://github.com/iconclass/data
  • Help documentation: https://iconclass.org/help/lod

Data Usage & Citation

Source Institution: Bibliotheca Hertziana - Max Planck Institute for Art History

  • Website: https://www.biblhertz.it/
  • Fotothek: https://fotothek.biblhertz.it/

AI Processing:

  • Model: Google Gemma 2 9B Instruct
  • Infrastructure: KISSKI (GWDG Göttingen)
  • Processing date: November 2024

ICONCLASS:

  • ICONCLASS classification system
  • Data source: https://github.com/iconclass/data

License: Please refer to the Bibliotheca Hertziana for source data licensing terms.

Quality Notes

General AI-Generated Content

  • AI-generated texts are meant to enhance discoverability and accessibility
  • Generated descriptions may contain inaccuracies or interpretations
  • Always refer to original structured metadata (textobj, textfoto) for authoritative information
  • Translations preserve meaning but may not capture all nuances of art historical terminology

Generated: November 2024 Processing location: KISSKI HPC Cluster, GWDG Göttingen Contact: pietro.liuzzo@biblhertz.it

Keyword Drawings
Related Publication Fotothek Kataloghttps://foto.biblhertz.it/search/fotos/encoded/eJzjYBKy4lIsT02KT8svyS-OT0tMTi2JL08tyk6NL05MzkhKTS_KTEsT4oxKzUzOyCvNS5didvRzUWIuycnWYgAAia8UTA
License/Data Use Agreement

Custom Dataset Terms

1 File
zeichnungen.zip
ZIP Archive - 437.4 MB
Published Nov 24, 2025
7 Downloads
MD5: 8e96093693df8bd3ee86f5ce3ecce9bb
Preview "zeichnungen.zip"
Edit File

This file has already been deleted (or replaced) in the current version. It may not be edited.

Restrict Access

Restricting limits access to published files. People who want to use the restricted files can request access by default. If you disable request access, you must add information about access to the Terms of Access field.

Learn about restricting files and dataset access in the User Guide.

Enable access request
You must enable request access or add terms of access to restrict file access.
Save Changes
Edit Embargo

The selected file or files have already been published. Contact an administrator to change the embargo date or reason of the file or files.
Edit Retention Period

The selected file or files have already been published. Contact an administrator to change the retention period date or reason of the file or files.
Delete Files

The file will be deleted after you click on the Delete button.

Files will not be removed from previously published versions of the dataset.

Select File(s)

Please select one or more files.

Share Dataset

Share this dataset on your favorite social media networks.

Continue
Dataset Citations

Citations for this dataset are retrieved from Crossref via DataCite using Make Data Count standards. For more information about dataset metrics, please refer to the User Guide.

Sorry, no citations were found.
Inaccessible Files Selected

The selected file(s) may not be downloaded because you have not been granted access or the file(s) have a retention period that has expired or the files can only be transferred via Globus.

You may request access to any restricted file(s) by clicking the Request Access button.

Ineligible Files Selected

The selected file(s) may not be transferred because you have not been granted access or the file(s) have a retention period that has expired or the files are not Globus accessible.

You may request access to any restricted file(s) by clicking the Request Access button.

Download Options

The files selected are too large to download as a ZIP.

You can select individual files that are below the 4.7 GB download limit from the files table, or use the Data Access API for programmatic access to the files.

Select File(s)

Please select a file or files to be downloaded.

Inaccessible Files Selected

The selected file(s) may not be downloaded because you have not been granted access or the file(s) have a retention period that has expired.

Click Continue to download the files you have access to download.

Ineligible Files Selected

Some file(s) cannot be transferred. (They are restricted, embargoed, with an expired retention period, or not Globus accessible.)

Click Continue to transfer the elligible files.

Delete Dataset

Are you sure you want to delete this dataset and all of its files? You cannot undelete this dataset.

Delete Draft Version

Are you sure you want to delete this draft version? Files will be reverted to the most recently published version. You cannot undelete this draft.

Unpublished Dataset Preview URL

Preview URL can only be used with unpublished versions of datasets.

Unpublished Dataset Preview URL

Are you sure you want to disable the Preview URL? If you have shared the Preview URL with others they will no longer be able to use it to access your unpublished dataset.

Delete Files

The file(s) will be deleted after you click on the Delete button.

Files will not be removed from previously published versions of the dataset.

Compute

This dataset contains restricted files you may not compute on because you have not been granted access.

Deaccession Dataset

Are you sure you want to deaccession? This is permanent and the selected version(s) will no longer be viewable by the public.

Deaccession Dataset

Are you sure you want to deaccession this dataset? This is permanent an it will no longer be viewable by the public.

Version Differences Details

Please select two versions to view the differences.

Version Differences Details
 
Version:
Last Updated:
Version:
Last Updated:
Select File(s)

Please select a file or files for access request.

Select File(s)

Embargoed files cannot be accessed. Please select an unembargoed file or files for your access request.

Edit Tags

Select existing file tags or create new tags to describe your files. Each file can have more than one tag.

Request Access

  You need to Sign Up or Log In to request access.

Dataset Terms

Please confirm and/or complete the information needed below in order to request access to files in this dataset.

This dataset is made available under the following terms. Please confirm and/or complete the information needed below in order to continue.

Our Community Norms as well as good scientific practices expect that proper credit is given via citation. Please use the data citation shown on the dataset page.

Custom Dataset Terms - the following Custom Dataset Terms have been defined for this dataset.

For copyright reasons the Fotothek database can only list digital images for which the Bibliotheca Hertziana, Max Planck Institute for Art History in Rome owns the copyright or the right of use. The contents and materials created by the site operators on these pages are subject to German copyright law. Duplication, processing, distribution, or any form of commercialization of such material beyond the scope of the copyright law shall require the prior written consent of its respective author or creator. Downloads and copies of this site are only permitted for private, non-commercial use. Insofar as the operator did not create the content on this site, the copyrights of third parties are respected. In particular, third-party content is identified as such. Should you nevertheless notice a copyright infringement, we request that you notify us accordingly. If we discover any infringements, we will remove such content immediately.
Preview Guestbook

Upon downloading files the guestbook asks for the following information.

Account Information

Package File Download

Use the Download URL in a Wget command or a download manager to download this package file. Download via web browser is not recommended. User Guide - Downloading a Dataverse Package via URL

https://edmond.mpg.de/api/access/datafile/

Compute Batch
Clear Batch
Dataset Persistent Identifier Change Compute Batch
Compute Batch
Submit for Review

You will not be able to make changes to this dataset while it is in review.

Publish Dataset

Are you sure you want to republish this dataset?

Make sure you have checked legal specifications for export control for academic publications by the German Federal Office for Economic Affairs and Export Control.

Select if this is a minor or major version update.

Publish Dataset

This dataset cannot be published until Edmond is published by its administrator.

Return to Author

Return this dataset to contributor for modification. The reason for return entered below will be sent by email to the author.

Curation Status History
StatusDateAssigner
No records found.
Add/Edit a Version Note
Styled Citation