Vision-based workface assessment using depth images for activity analysis of interior construction operations

Workface assessment –the process of determining the overall activity rates of onsite construction workers throughout a day– typically involves manual visual observations which are time-consuming and labor-intensive. To minimize subjectivity and the time required for conducting detailed assessments, and allowing managers to spend their time on the more important task of assessing and implementing improvements, we propose a new inexpensive vision-based method using RGB-D sensors that is applicable to interior construction operations. This is particularly a challenging task as construction activities have a large range of intra-class variability including varying sequences of body posture and time-spent on each individual activity. On the other hand, the state-of-the-art skeleton extraction algorithms from RGB-D sequences are not robust enough especially when workers interact with tools or self-occlude the camera’s field-of-view. Existing vision-based methods are also rather limited as they can primarily classify “atomic” activities from RGB-D sequences involving one worker conducting a single activity. To address these limitations, our proposed original method involves three main components: 1) an algorithm for detecting, tracking, and extracting body skeleton features from depth images; 2) A discriminative bag-of-poses activity classifier trained using multiple Support Vector Machines for classifying single visual activities from a given body skeleton sequence; and 3) a Hidden Markov model with a Kernel Density Estimation function to represent emission probabilities in form of a statistical distribution of single activity classifiers. For training and testing purposes, we also introduce a new dataset of eleven RGB-D sequences for interior drywall construction operations involving three actual construction workers conducting eight different activities in various interior locations. Our experimental results with an average accuracy of 76% on the testing dataset show the promise of vision-based methods using RGB-D sequences for facilitating the activity analysis workface assessment