at
| @silkiways The work seems to use trained descriptors in order to distinguish parts of shapes. This could for example be used in shape queries. |
| — Posted to eurographics 2011 at 12:56, April 14, 2011 |