SEMANTIC SEGMENTATION OF AERIAL IMAGES IN URBAN AREAS WITH CLASS-SPECIFIC HIGHER-ORDER CLIQUES

By Javier A. Montoya-Zegarra, Jan Dirk Wegner, Lubor Ladicky, Konrad Schindler. 2015

53 citations recorded by the source corpus.

Abstract. In this paper we propose an approach to multi-class semantic segmentation of urban areas in high-resolution aerial images with classspecific object priors for buildings and roads. What makes model design challenging are highly heterogeneous object appearances and shapes that call for priors beyond standard smoothness or co-occurrence assumptions. The data term of our energy function consists of a pixel-wise classifier that learns local co-occurrence patterns in urban environments. To specifically model the structure of roads and buildings, we add high-level shape representations for both classes by sampling large sets of putative object candidates. Buildings are represented by sets of compact polygons, while roads are modeled as a collection of long, narrow segments. To obtain the final pixel-wise labeling, we use a CRF with higher-order potentials that balances the data term with the object candidates. We achieve overall labeling accuracies of > 80%.

Subfields: Computer Vision and Pattern Recognition, Environmental Engineering, Ocean Engineering, Remote Sensing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Andreas Geiger
2. Fatma Güney
3. Joel Janai
4. Aseem Behl
5. Konrad Schindler
6. Luc Van Gool
7. Silvano Galliani
8. Jitendra Malik

In Environmental Engineering

1. Fatma Güney
2. Joel Janai
3. Konrad Schindler
4. Jan Dirk Wegner
5. Mihai Datcu
6. Uwe Stilla
7. Aseem Behl
8. Silvano Galliani

In Ocean Engineering

1. Konrad Schindler
2. Jan Dirk Wegner
3. Mihai Datcu
4. Uwe Stilla
5. Silvano Galliani
6. Dimitrios Marmanis
7. Fatma Güney
8. Andreas Geiger

In Remote Sensing

1. Mihai Datcu
2. Liangliang Nan

Citations in the corpus

• 3D Instance Segmentation of MVS Buildings (2021, methodology): “use an α-shape algorithm to calculate the boundary polygons of building objects, which are further optimized by conditional random field (CRF) [32].”
• SEMANTIC SEGMENTATION OF AERIAL IMAGES WITH AN ENSEMBLE OF CNNS (2016, background): “…put on including a-priori information about the class layout like smoothness, shape templates, and long-range connectivity [Karantzalos and Paragios, 2009, Lafarge et al., 2010, Schindler, 2012, Montoya-Zegarra et al., 2015], often in the form of Conditional Random Fields or Marked Point Processes.”
• Computer Vision for Autonomous Vehicles: Problems, Datasets and State-of-the-Art (2017, background): “Montoya et al. (2015) extend this formulation to multi-label classification of aerial images with class-specific priors for buildings and roads.”