Datensatz

Zusammenfassung

<p>Recognition of object classes in natural images has made tremendous\u000A progress in recent years. Today’s approaches often\u000A rely on powerful learning approaches as well as robust local 2D shape\u000A or appearance features. Exploiting 3D shape cues however has become\u000A unfashionable in recent literature. While shading cues play a major\u000A role in human perception of object shape, shape‐from‐shading techniques\u000A are seldom used today for object class detection. Drawing on ideas\u000A from the early days in object recognition this paper aims to revisit\u000A the concept of using shading primitives to support object class detection.\u000A We demonstrate and discuss the applicability of this approach to\u000A real world images of a standard benchmark data set. Experimental\u000A results suggest that our shading cues can be useful for object class\u000A detection.</p>