Recently, 3D data generation techniques using neural networks have been widely used across various fields. In this study, we propose an efficient interactive annotation system that combines high-quality 3D scene models generated by 3D Gaussian Splatting with models that have surface information, such as mesh models. This system allows users to intuitively annotate 3D spaces through interactive operations. Our system promotes the effective use and sharing of 3D data, making it applicable in various domains. In particular, it is expected to be highly beneficial for disaster response, where rapid assessment of affected areas is crucial. By generating 3D models from aerial images captured by drones, the system enables detailed visualization of structural damage, terrain changes, and road blockages. This information helps rescue teams efficiently determine passable routes and optimize relief supply transportation plans. Furthermore, sharing highly accurate 3D representations of disaster-affected areas with relevant authorities in real time strengthens coordination among emergency response teams.