We are developing a support system for maintenance training of electric power facilities by augmented reality. A user of the system wears a Head-Mounted Display (HMD) with a small camera, puts a sensor on top of one?s head and stands in front of a blue screen. The system recognizes an object and estimates its position and pose from an image that is captured by the small camera. The object recognition and the position-pose estimation are performed by image processing without any special image marks or sensors. After the object recognition and the position-pose estimation, the system overlays computer graphics to the object images. The computer graphics show operational guidance, instructions, or dynamic inside movements of the object on the HMD. In this paper, we describe an outline of the system and evaluation results of the system functions: the object recognition, the position-pose estimation and the operational guidance.