Images of Construction Equipment

A dataset when it comes to recognition of excavator, loader, dozer, roller and backhoe

An MATLAB-based annotation tool when it comes to annotation of construction web site photos

The comparison of recognition practices regarding correctness, robustness and speed

Tips for choosing appropriate recognition practices

The recognition of construction functional sources (equipment, workers, products, etc.) features played an important role in attaining completely automated building. So far, many item recognition practices happen developed in computer sight; however, they have been tested with some categories of objects in all-natural moments. Consequently, their particular performance in the recognition of construction working sources is not clear, specifically thinking about building web sites are generally dirty, disorderly, and cluttered. This paper proposes a typical dataset of construction web site pictures determine the building gear recognition overall performance of current item recognition methods. A large number of pictures have-been collected and created, which cover 5 courses of construction equipment (excavator, loader, dozer, roller and backhoe). Each image happens to be annotated aided by the equipment type, location, positioning, occlusion, and labeling of equipment elements (bucket, stick, growth etc.). The potency of the dataset has-been examined with two well-known object recognition techniques in computer vision. The outcomes reveal the dataset could effectively determine the performance among these methods when it comes to correctness, robustness, and speed of acknowledging construction gear.

Keywords

• Automation;
• Building equipment;
• Dataset;
• Evaluation;
• Efficiency attributes