Object Proposal Using RGB-D Edge Density and Density Uniformity

DOI

Bibliographic Information

Other Title
  • RGB-D画像のエッジの密度と均一性を用いたObject Proposal

Description

Object proposal that detects candidate region of objects in an image is widely used in combination with object detection. The window scoring based methods usually have higher efficiency than object region extraction methods. However, the previous methods detect an excessive number of boxes due to edges of texture objects and design of objectness. To tackle these problems, we propose a novel objectness measure using an RGB-D image based on an assumption that an object box should circumscribes an object boundary from an object. Our experiments show significant performance gains over existing RGB and RGB-D object proposal methods on multiple dataset.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top