MBDD2025: UAV Images of Building Wall Defects

MBDD2025 contains 14,471 UAV images of building wall defects. The images are annotated in the VOC format and released under a CC BY 4.0 license. The dataset is hosted on Kaggle.

Use Cases

• Train object detection models for building defect identification based on UAV imagery.
• Benchmark defect detection algorithms using VOC-format annotations.
• Develop automated inspection systems for construction and maintenance based on aerial wall images.

Strengths

• Contains 14,471 annotated images, providing a substantial base for model training.
• Annotations follow the VOC format, a widely supported standard for object detection tasks.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or temporal bias inherent to the unspecified collection source.

Provenance

Collection Method

Likely collected via UAV (drone) photography of building exteriors.