Toward Automatic 3D Generic Object Modeling from One Single Image

We present a novel method for solving the challenging problem of generating 3D models of generic object categories from just one single un-calibrated image. Our method leverages the algorithm proposed in [44] which enables a partial reconstruction of the object from a single view. A full reconstruction is achieved in an object completion stage where modified or state-of-the-art 3D shape and texture completions techniques are used to recover the complete 3D model. We present results of our method on a number of images containing objects from five generic categories (mice, staplers, mugs, cars, and bicycles). We demonstrate (numerically and quantitatively) that our method produces convincing 3D models from a single image using minimal or no human intervention. Our technique is targeted to applications where users are interested in building virtual collections of 3D models of objects, and sharing such models in virtual environments such as Google 3D Warehouse or Second Life (secondlife.com).