Abstract:Currently most of the campus displays are developed through virtual reality technologyꎬ which gets users im ̄ mersed in the virtual worldꎬ but ignore information of the real world. In this paperꎬ a campus display application is de ̄ signed based on Unity3D+ARCore+TensorFlowꎬ which provide functions of plane detectionꎬ multi-image and object rec ̄ ognition. The implementation process and problems of virtual objectsꎬ touch screen interactionꎬ plane detectionꎬ image and object recognition are described in detail. Performance optimization methods in scripting and UI design are illustrated. The test shows that multi-marker recognition improves practicabilityꎬ and object recognition enriches application implicationsꎬ which together could increase the design methods of campus displays and expand the application scope of augmented reality technology.