A merged dataset combines the RDD-2022 dataset from six countries with two supplementary aerial datasets, UAV-PDD2023 and RoadDamageVision from China and Spain. It is a derived dataset repackaged and relabeled by TamAko783 from three independently published sources. The dataset was last updated on June 4, 2026.
Use Cases
- Training object detection models for road defect classification based on the unified 4-class CRDDC schema.
- Benchmarking model performance on combined ground-level and aerial/drone imagery.
- Developing infrastructure inspection systems using multi-source visual data.
Strengths
- Combines data from three distinct sources: RDD-2022, UAV-PDD2023, and RoadDamageVision.
- Unifies annotations into a single 4-class schema for consistency.
- Includes imagery from both ground-level and aerial/drone perspectives.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- Derived from three independently published sources: RDD-2022, UAV-PDD2023, and RoadDamageVision.
- Collection Method
- Repackaged and relabeled into a single YOLO-format dataset.
- Freshness
- Last updated 2026-06-04 07:10:26; freshness should be verified.
- Geography
- Likely includes data from six countries (RDD-2022) and China and Spain (supplementary datasets).