en
0.25
0.5
0.75
1.25
1.5
1.75
2
A Three-Layered Approach to Facade Parsing
Published on Nov 12, 20123852 Views
We propose a novel three-layered approach for semantic segmentation of building facades. In the first layer, starting from an oversegmentation of a facade, we employ the recently introduced machine le
Related categories
Chapter list
A Three-Layered Approach to Facade Parsing00:00
We aim to improve the state of the art in facade parsing00:04
We do not use shape grammars!00:35
A Three-Layered Approach01:51
Bottom layer - segments02:33
Image preparation02:38
Recursive Neural Network02:55
Bottom Layer Output03:42
Middle layer - objects04:05
Window and Door Detection04:13
Incorporating Detector Knowledge With MRFs - 104:31
Incorporating Detector Knowledge With MRFs - 204:42
Incorporating Detector Knowledge With MRFs - 304:50
From Bottom To Middle Layer Output05:03
Top layer - architectural elements05:35
Weak Architectural Principles05:40
From Middle To Top Layer Output07:07
Ecole Centrale Paris Facades Database - 107:37
Ecole Centrale Paris Facades Database - 207:56
Ecole Centrale Paris Facades Database - 308:11
Results - ECP Dataset - 108:37
Pixel Accuracy vs Visual Effect09:05
Results - ECP Dataset - 209:36
Example Outputs - ECP Dataset09:48
eTRIMS Database10:01
Example Outputs - eTRIMS Dataset10:18
Example Outputs - Procedural Models10:29
Summary11:01
Questions?11:40