Abstract

Category-level ob ject detection, the task of locating ob ject instances of a given category in images, has been tackled with many al- gorithms employing standard color images. Less attention has been given to solving it using range and depth data, which has lately become read- ily available using laser and RGB-D cameras. Exploiting the different nature of the depth modality, we propose a novel shape-based ob ject detector with partial pose estimation for axial or reflection symmetric ob jects. We estimate this partial pose by detecting target’s symmetry, which as a global mid-level feature provides us with a robust frame of reference with which shape features are represented for detection. Re- sults are shown on a particularly challenging depth dataset and exhibit significant improvement compared to the prior art.