This paper describes a novel view-based algorithm for 3D object recognition using a network of linear units. The SNoW learning architecture is a sparse network of linear functions over a pre-defined or incrementally learned feature space and is specifically tailored for learning in the presence of a very large number of features. We use the pixel-level representation in the experiments and compare the performance of SNoW with Support Vector Machines and nearest neighbor methods on 3D object recognition using the 100 objects in the Columbia Image Object Database (COIL-100). Experimental results show that SNoW-based method outperform SVM-based system in terms of recognition rate and the computational cost involved in learning. The empirical results also provide insight for practical and theoretical considerations on view-based methods for 3D object recognition.
Back to publication home page