We present a data-driven approach that colorizes 3D furniture models andindoor scenes by leveraging indoor images on the internet. Our approach is ableto colorize the furniture automatically according to an example image. The coreis to learn image-guided mesh segmentation to segment the model into differentparts according to the image object. Given an indoor scene, the system supportscolorization-by-example, and has the ability to recommend the colorizationscheme that is consistent with a user-desired color theme. The latter isrealized by formulating the problem as a Markov random field model that imposesuser input as an additional constraint. We contribute to the community ahierarchically organized image-model database with correspondences between eachimage and the corresponding model at the part-level. Our experiments and a userstudy show that our system produces perceptually convincing results comparableto those generated by interior designers.
展开▼
