泡泡一分鐘:SceneCut: Joint Geometric and Object Segmentation for Indoor Scenes

張寧    SceneCut: Joint Geometric and Object Segmentation for Indoor Scenes
    "連接：https://pan.baidu.com/s/1hpwb8IjtEpb3uTIncJbTUg
提取碼：0wz1"

Trung T. Pham  , Thanh-Toan Do  , Niko Sünderhauf , Ian Reid

SceneCut：室內場景的聯合幾何和對象分割

This paper presents SceneCut, a novel approach to jointly discover previously unseen objects and non-object surfaces using a single RGB-D image. SceneCut’s joint reasoning over scene semantics and geometry allows a robot to detect and segment object instances in complex scenes where modern deep learning-based methods either fail to separate object instances, or fail to detect objects that were not seen during training. SceneCut automatically decomposes a scene into meaningful regions which either represent objects or scene surfaces. The decomposition is qualified by an unified energy function over objectness and geometric fitting. We show how this energy function can be optimized efficiently by utilizing hierarchical segmentation trees. Moreover, we leverage a pretrained convolutional oriented boundary network to predict accurate boundaries from images, which are used to construct high-quality region hierarchies. We evaluate SceneCut on several different indoor environments, and the results show that SceneCut significantly outperforms all the existing methods.

本文介紹了SceneCut，這是一種使用單個RGB-D圖像聯合發現之前看不見的物體和非物體表面的新方法。SceneCut對場景語義和幾何的聯合推理容許機器人在複雜場景中檢測和分割對象實例，其中現代基於深度學習的方法沒法分離對象實例，或者沒法檢測在訓練期間未看到的對象。SceneCut會自動將場景分解爲有意義的區域，這些區域表明對象或場景表面。經過對象性和幾何擬合的統一能量函數來限定分解。咱們展現瞭如何經過利用分層分割樹有效地優化這種能量函數。此外，咱們利用預訓練的卷積導向邊界網絡來預測圖像的準確邊界，這些邊界用於構建高質量的區域層次結構。咱們在幾個不一樣的室內環境中評估SceneCut，結果代表SceneCut明顯優於全部現有方法。