A skill is important factor integrate technology of machine and technique of human. These determine the quality level of the defect inspection process. Panels with advanced skills, have an ability to perform the actions that meet state and powers of observation to work on, the state of the work object. An important sense is concerned with a perceptual and cognitive ability. It should hurry to develop a method to quantitatively assess the characteristics of human perception. In this paper, in defect detection experimental evaluation of panels based on the inspection image for fracture surface of metallic materials, we focus on the relationship of non-defective rejection probability and defect detection probability from the probabilistic model based on signal detection theory, for the characterization of panels Experiments were carried out to calculate the metrics according to the skill level of panels that conform to the flow of international standardization in recent years. Perception of the panel, is represented by ROC curve, skill levels showed that the degree of knowledge and experience can be categorized quantitatively.