Part of Advances in Neural Information Processing Systems 11 (NIPS 1998)
Sergey Ioffe, David Forsyth
Finding articulated objects, like people, in pictures present.s a par(cid:173) ticularly difficult object. recognition problem. We show how t.o find people by finding putative body segments, and then construct.(cid:173) ing assemblies of those segments that are consist.ent with the con(cid:173) straints on the appearance of a person that result from kinematic properties. Since a reasonable model of a person requires at. least nine segments, it is not possible to present every group to a classi(cid:173) fier. Instead, the search can be pruned by using projected versions of a classifier that accepts groups corresponding to people. We describe an efficient projection algorithm for one popular classi(cid:173) fier , and demonstrate that our approach can be used to determine whether images of real scenes contain people.