Nvidia Hackathon Dataset: A Curated Index for Crack and Pavement Segmentation

A curated reference index of public datasets for crack and pavement segmentation tasks. The repository points to canonical sources for images and masks, enabling researchers to download data under original licenses from the original authors. The index is maintained by user 'crackedcity' and was last updated on May 31, 2026.

Use Cases

• Training semantic segmentation models based on the described crack and pavement imagery.
• Benchmarking model performance across different public datasets referenced in the index.
• Conducting meta-analysis on available public data sources for infrastructure inspection.
• Developing automated pavement condition assessment tools based on the linked image datasets.

Strengths

• Indexes multiple public datasets from canonical sources, facilitating discovery.
• Provides metadata such as publisher, source link, and approximate size for each referenced dataset.
• Last update timestamp (2026-05-31) is provided, indicating recent maintenance.

Limitations

• The dataset itself contains no images, only metadata and links; actual data quality and scale depend on external sources.
• Column-level documentation for the index is absent; field semantics must be inferred.
• Row count and specific file formats for the index are unknown, limiting suitability assessment.

Provenance

Source

huggingface

Collection Method

Curated compilation of references to third-party public datasets.

Freshness

Last updated 2026-05-31 14:45:40.