This presentation introduces indirect sensing in cyber-physical systems that infers desired information using structures as sensors (through physical structural vibration responses) instead of directly measuring sensing targets. Smart structures are designed to sense, understand, and respond to various situations involving structure itself, humans, and surrounding environment, to enhance user safety, productivity, and quality of life. Current approaches involve dense deployment of dedicated sensors, which makes large scale monitoring challenging. Advantages of indirect monitoring are 1) multiple types of target information can be obtained by sensing only structural responses; 2) sparse sensing configuration is possible by utilizing structural properties; and 3) privacy is better preserved due to the hidden and indirect nature of the approach. The structures of interest include buildings and vehicles. In buildings, various information on human behavior (e.g., location, height, weight, etc.), building systems (structural health, heating systems, etc.), and environment (e.g., traffic, weather, etc.) can be indirectly inferred from high-resolution, high-frequency structural vibration data. In vehicles, we can diagnose the vehicle, infrastructures, and humans based on vehicular vibration data. Through our collaboration with government and industry partners, we have deployed several buildings and train systems to validate and apply our research.