Effective Building Extraction by Learning to Detect and Correct Erroneous Labels in Segmentation Mask - École des Ponts ParisTech Access content directly
Conference Papers Year : 2018

Effective Building Extraction by Learning to Detect and Correct Erroneous Labels in Segmentation Mask

Nikos Komodakis
  • Function : Author
  • PersonId : 945099

Abstract

Semantic segmentation is pivotal for remote sensing image analysis. Although existing segmentation techniques perform well on similar landscape images, their generalization capability on an entirely different landscape is extremely poor. One of the primary reasons is that they partially or wholly, neglect the underlying relationship that exist in the joint space of input and output variables. Thus, effectively they lack to impose structure in their output predictions which is necessary for successful segmentation. In this paper, we address this problem and propose a novel solution by modeling the joint distribution of input-output variable which in turn enforces some structure in the initial segmentation mask. To this end, we first detect erroneous labels, in the form of Error maps, in the initial building masks. These Error maps are then used to correct the corresponding erroneous labels through a replacement technique. We evaluate our methodology on the benchmark Inria Aerial Image Labeling dataset, which is a large scale high resolution dataset for building footprint seg-mentation. In contrast to previous methods, our predicted seg-mentation masks are much closer to ground truth, owning to the fact that they are able to effectively correct both the large errors as well as the blobby effects. We lastly perform on par with other state-of-the-arts, validating the efficacy of our technique.
Fichier principal
Vignette du fichier
singh.pdf (4.34 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01832798 , version 1 (11-08-2018)

Identifiers

  • HAL Id : hal-01832798 , version 1

Cite

Praveer Singh, Nikos Komodakis. Effective Building Extraction by Learning to Detect and Correct Erroneous Labels in Segmentation Mask. IGARSS, 2018, Valencia, Spain. ⟨hal-01832798⟩
74 View
180 Download

Share

Gmail Facebook X LinkedIn More