Industry 5.0 aims to shift the production paradigm from system-centric to human-centric. Human-centric production requires interoperability between humans and technical systems. Human behavior analysis is one of the keys to contributing to interoperability. Current approaches analyze general behavior based on databases, but consideration for individual behavior is lacking. The critical challenge to analyzing an individual behavior is behavior labeling. Therefore, this paper aims to propose an approach to label individual behavior in industrial environments to enhance the interoperability of humans and technical systems. The approach applies feature generation, low-dimensional embedding, and segmentation to label individual behavior. A case study of power transformer assembly is utilized to demonstrate the feasibility of the approach. The results show that human behavior can be embedded into 2D dimensions, and the behavior tends to accumulate in distinct areas. The distinct areas are segmented to yield the behavior labels. This approach is suitable for analyzing human behavior in industrial environments. This approach could be implemented in human–robot collaboration tasks for adaptive robot control to achieve the interoperability of humans and technical systems.