Home
Uni-Logo
 

Pixel-level encoding and depth layering for instance-level semantic labeling

Jonas Uhrig, M. Cordts, U. Franke, Thomas Brox
German Conference on Pattern Recognition (GCPR), 2016
Abstract: Recent approaches for instance-aware semantic labeling have augmented convolutional neural networks (CNNs) with complex multi-task architectures or computationally expensive graphical models. We present a method that leverages a fully convolutional network (FCN) to predict semantic labels, depth and an instance-based encoding using each pixel's direction towards its corresponding instance center. Subsqquently, we apply low-level computer vision techniques to generate state-of-the-art instance segmentation on the street scene datasets KITTI and Cityscapes. Our approach outperforms existing works by a large margin and can additionally predict absolute distances of individual instances from a monocular image as well as a pixel-level semantic labeling.


Other associated files : uhrig_gcpr16.pdf [4.4MB]  

Images and movies

 

BibTex reference

@InProceedings{BU16,
  author       = "J. Uhrig and M. Cordts and U. Franke and T. Brox",
  title        = "Pixel-level encoding and depth layering for instance-level semantic labeling",
  booktitle    = "German Conference on Pattern Recognition (GCPR)",
  month        = " ",
  year         = "2016",
  url          = "http://lmb.informatik.uni-freiburg.de/Publications/2016/BU16"
}

Other publications in the database