Construction Safety Object Detection Images with 17 Safety-Related Labels

398 images split into training (307), validation (57), and test (34) sets for object detection tasks. The dataset, created by keremberke and last updated on January 27, 2023, is annotated with labels for safety equipment and violations. It includes classes such as 'hardhat', 'safety vest', 'no-hardhat', 'person', and construction vehicles.

Use Cases

• Training object detection models to identify personal protective equipment (PPE) like hardhats and safety vests based on the listed labels.
• Developing systems to detect safety violations such as 'no-hardhat' or 'no-safety vest' from site imagery.
• Benchmarking model performance on detecting construction machinery like excavators, dump trucks, and wheel loaders.
• Creating automated site safety audit tools by classifying objects like barricades, safety nets, and gloves.

Strengths

• Provides a clear train/validation/test split with 307, 57, and 34 images respectively.
• Focuses on a specific, practical domain with 17 distinct object classes relevant to construction safety.
• Hosted on a major platform (Hugging Face) with a specific last update timestamp of 2023-01-27 13:36:19.

Limitations

• The total number of images (398) is relatively small for deep learning tasks.
• Column-level documentation, file formats, and dataset size details are unknown, requiring inspection after download.
• Geographic origin, collection method, and temporal coverage of the images are not specified, which may indicate unknown biases.

Provenance

Source

keremberke on Hugging Face.

Freshness

Last updated 2023-01-27 13:36:19.