Physical exercise instruction sheets are often difficult to understand. In most cases considerable information remains implicit, which also poses a considerable hurdle regarding potential machine understanding. Major missing information types include the source and destination location of a human movement. We present a Bayesian network for extracting the implicit or missing information from typical exercise instruction sheets. We propose two different kinds of Bayesian networks which consist of three and four variables respectively. The network with three variables is designed for single exercise instruction, featuring a single action or pose. The other is designed for single of multiple sentence(s) spanning two actions or poses. The conditional probability table (CPT) is the backbone of a Bayesian network. At the start, the CPT is informed by our physical exercise instruction sheet corpus (PEISC). Keeping the Action and Bodypart fixed, we have developed our CPT using a crowdsourcing. We have developed a CPT update system using 13 different exercises consisting of 44 different exercise videos. Using this system, candidate exercise executions are rated by participants and their ratings update the CPT. We also update the Action variable, which consists of 14 different values (action verbs) using crowdsourcing with a human computation approach.