An approach to define construction practice from construction images

In the construction industry, safety is an important, but often neglected, factor. Due to its nature construction is one of the most hazardous industries and so identifying safe construction practices becomes very important. Nowadays, the use of construction images has become very popular. As information sources, images/pictures recorded of the construction process provide significant information relating to the safe construction practice. However, there are a number of problems related to using images to determine if the construction practices being used are safe or unsafe because of uncertain or inexact information collected by looking at that image. To deal with the uncertain information from construction images, it is proposed to use a database of images alongside their characteristics and an assessment [made by direct assessment on-site] of their safety. The framework proposed for this is described. It is initially based on Bayes' Theorem. From 20 construction images as an example, the result revealed that eight images demonstrate safe construction practices, whereas 12 images demonstrate unsafe construction practices. In conclusion, proposed approach demonstrates how it is possible, using on-site images, to define construction practice as safe or unsafe.