AOW-YOLO: Annotated Images for Smoking Detection on Construction Sites

AOW-YOLO is a novel model designed to detect smoking behavior in complex construction site environments, where objects are typically small and backgrounds cluttered. The model, proposed by Ruishi Liang, reportedly surpasses existing lightweight models in mAP metrics while achieving roughly 31.6% faster inference speed than YOLO11n. The underlying dataset, shared on figshare under a CC-BY-4.0 license, was last updated in May 2026.

Use Cases

• Training object detection models for safety compliance based on images of small objects in cluttered environments.
• Benchmarking lightweight detection models based on the reported mAP50 and mAP50:95 metrics.
• Developing construction site safety management applications based on the described detection of smoking behavior.

Strengths

• Model performance is quantified with specific mAP50 and mAP50:95 metrics.
• The associated model is reported to achieve roughly 31.6% faster inference speed than a baseline.

Limitations

• Dataset size is only 5.5 KB, indicating a very limited scope of data.
• Row count and column-level documentation are unknown, which may limit suitability assessment.
• The description focuses on the model architecture; the nature and scale of the underlying image data are not detailed.

Provenance

Source

Ruishi Liang

Collection Method

Likely contains images collected or annotated for model training, as described in the research summary.

Freshness

Last updated 2026-05-06 17:45:42.