ShapeDomainLab: A Controlled Synthetic Dataset of Simple Geometric Shapes for Studying Domain Robustness of Computer Vision Models (doi:10.48788/DVUA/T8V83V)
(ShapeDomainLab: контрольований синтетичний датасет простих геометричних фігур для дослідження доменної стійкості моделей комп’ютерного зору)

View:

Part 1: Document Description
Part 2: Study Description
Part 3: Data Files Description
Part 4: Variable Description
Part 5: Other Study-Related Materials
Entire Codebook

ShapeDomainLab: A Controlled Synthetic Dataset of Simple Geometric Shapes for Studying Domain Robustness of Computer Vision Models

doi:10.48788/DVUA/T8V83V

DataverseUA

2026-08-13

1

Galelyuka, Igor; Romanov, Volodymyr, 2026, "ShapeDomainLab: A Controlled Synthetic Dataset of Simple Geometric Shapes for Studying Domain Robustness of Computer Vision Models", https://doi.org/10.48788/DVUA/T8V83V, DataverseUA, V1, UNF:6:bqdw6+LCffW4aqx10ClhDA== [fileUNF]

ShapeDomainLab: A Controlled Synthetic Dataset of Simple Geometric Shapes for Studying Domain Robustness of Computer Vision Models

ShapeDomainLab: контрольований синтетичний датасет простих геометричних фігур для дослідження доменної стійкості моделей комп’ютерного зору

doi:10.48788/DVUA/T8V83V

Galelyuka, Igor (V.M. Glushkov Institute of Cybernetics of the National Academy of Sciences of Ukraine)

Romanov, Volodymyr (V.M. Glushkov Institute of Cybernetics of the National Academy of Sciences of Ukraine)

ShapeDomainLab (Shape Domain Laboratory)

DataverseUA

Galelyuka, Igor

Galelyuka, Igor

2026-08-11

https://doi.org/10.48788/DVUA/T8V83V

Computer and Information Science, synthetic data, geometric shapes, computer vision, domain shift, domain robustness, segmentation, reproducibility

<b>## Motivation and scope</b><br> Motivation: controlled study of domain shift, domain robustness, classification and segmentation; this is not a substitute for natural data.<br><br> <b>## Composition</b><br> Six domains (clean, rotation, geometry, noise, clipping, composite); five classes (circle, octagon, rectangle, square, triangle); 51,000 64*64 grayscale PNG images and 51,000 binary L-mode PNG masks. Per domain: 5,000 train, 1,000 validation and 2,500 test images.<br><br> <b>## Generation and splits</b><br> Images are reproducibly generated controlled 2D shape rasters. Metadata schema version 2.1 records rendering parameters and file linkage. Train/validation/test splits are independent within each domain.<br><br> <b>## Intended uses and limitations</b><br> Suitable for controlled classification, segmentation, invariance and educational experiments. It has no natural textures, camera model, complex scenes or external validity for natural imagery. Strong clipping can reduce semantic identifiability.

Exact SHA-256 screening found 143 cross-domain pixel-identical groups: 41 expected parameter collisions and 102 expected symmetry cases; no group has different class labels. Fifty groups span train/test across domains. Strict cross-domain benchmarks must use CROSS_DOMAIN_DUPLICATES.csv as an exclusion/filter list and report the applied rule.

• Number of cases: 36

• No. of variables per record: 9

• Type of File: text/tab-separated-values

UNF:6:4IdasnXTHK7Yi2vg1HNMcA==

• Number of cases: 20

• No. of variables per record: 8

• Type of File: text/tab-separated-values

UNF:6:zVEd6EIHSgRQhfRQyqdMCQ==

• field_name - field_name
• data_type - data_type
• required - required
• unit - unit
• allowed_values - allowed_values
• description_ua - description_ua
• description_en - description_en
• example - example
• notes - notes
• field - field
• clean - clean
• rotation - rotation
• geometry - geometry
• noise - noise
• clipping - clipping
• composite - composite
• consistent - consistent

f1547 Location:

Variable Format: character

Notes: UNF:6:v/+8fDsnef2gdtFiJJAJxA==

f1547 Location:

Variable Format: character

Notes: UNF:6:BZSo010zX0IgkqHuAo+48g==

f1547 Location:

Variable Format: character

Notes: UNF:6:WtB+KpeM3zJmerpeXTDZGA==

f1547 Location:

Summary Statistics: Max. NaN; Mean NaN; Valid 0.0; Min. NaN; StDev NaN;

Variable Format: numeric

Notes: UNF:6:dxM/Qx1eEt2FAALA09Tg/g==

f1547 Location:

Summary Statistics: Valid 0.0; Min. NaN; StDev NaN; Max. NaN; Mean NaN;

Variable Format: numeric

Notes: UNF:6:dxM/Qx1eEt2FAALA09Tg/g==

f1547 Location:

Variable Format: character

Notes: UNF:6:xi3pu2e1OYGol1DriZ1mWw==

f1547 Location:

Variable Format: character

Notes: UNF:6:HBrbGQqlSyfKR3dSXm+PPA==

f1547 Location:

Summary Statistics: Mean NaN; Min. NaN; Valid 0.0; Max. NaN; StDev NaN;

Variable Format: numeric

Notes: UNF:6:dxM/Qx1eEt2FAALA09Tg/g==

f1547 Location:

Variable Format: character

Notes: UNF:6:gpb4iyZ8pN+QQhL20EuvlQ==

f1520 Location:

Variable Format: character

Notes: UNF:6:yJ6u1JuFk9TqNF2HCPPVEw==

f1520 Location:

Variable Format: character

Notes: UNF:6:it2LjA1YydCk7PNCIE+P3g==

f1520 Location:

Variable Format: character

Notes: UNF:6:0zNypawCLihGpdzmqjbN9g==

f1520 Location:

Variable Format: character

Notes: UNF:6:I4a0Kc+xLKTv3wypFz0QuA==

f1520 Location:

Variable Format: character

Notes: UNF:6:3ajOXGQDN9hMrQrJ4xcJDA==

f1520 Location:

Variable Format: character

Notes: UNF:6:ZfQhbS3VVDABN4FXJkMPww==

f1520 Location:

Variable Format: character

Notes: UNF:6:dUhaVZIu9n2AquxpYh2f/A==

f1520 Location:

Variable Format: character

Notes: UNF:6:NlNIXiy8IQunJJUCzXrYsg==

Label:

AUTHORS.md

text/markdown

Label:

CHANGELOG.md

text/markdown

Label:

CITATION.cff

application/octet-stream

Label:

CITATION.txt

text/plain

Label:

DATASET_CARD_EN.md

text/markdown

Label:

DATASET_CARD_UA.md

text/markdown

Label:

DATAVERSE_METADATA_EN.md

text/markdown

Label:

DATAVERSE_METADATA_UA.md

text/markdown

Label:

DOMAIN_SPECIFICATIONS_EN.md

text/markdown

Label:

DOMAIN_SPECIFICATIONS_UA.md

text/markdown

Label:

LICENSE_DATASET.txt

text/plain

Label:

LICENSE_STATUS.md

text/markdown

Label:

METADATA_DICTIONARY_EN.md

text/markdown

Label:

METADATA_DICTIONARY_UA.md

text/markdown

Label:

QUALITY_ASSURANCE.md

text/markdown

Label:

README_EN.md

text/markdown

Label:

README_UA.md

text/markdown

Label:

REPRODUCIBILITY.md

text/markdown

Label:

shapedomainlab_v1.0.0_analysis_reports.zip

application/zip

Label:

shapedomainlab_v1.0.0_clean.zip

application/zip

Label:

shapedomainlab_v1.0.0_clipping.zip

application/zip

Label:

shapedomainlab_v1.0.0_composite.zip

application/zip

Label:

shapedomainlab_v1.0.0_geometry.zip

application/zip

Label:

shapedomainlab_v1.0.0_noise.zip

application/zip

Label:

shapedomainlab_v1.0.0_quality_reports.zip

application/zip

Label:

shapedomainlab_v1.0.0_rotation.zip

application/zip

Label:

VERSION.txt

text/plain

Label:

clean.json

application/json

Label:

clipping.json

application/json

Label:

composite.json

application/json

Label:

geometry.json

application/json

Label:

noise.json

application/json

Label:

rotation.json

application/json

Label:

clean.json

application/json

Label:

clipping.json

application/json

Label:

composite.json

application/json

Label:

geometry.json

application/json

Label:

noise.json

application/json

Label:

rotation.json

application/json

Label:

environment.json

application/json

Label:

git_branch.txt

text/plain

Label:

git_head.txt

text/plain

Label:

git_status_before.txt

text/plain

Label:

pip_list.json

application/json

Label:

PREFLIGHT_REPORT.md

text/markdown

Label:

test_results.txt

text/plain