IEEE Geoscience and Remote Sensing Magazine - March 2023 - 94

TECHNICAL COMMITTEES
CLAUDIO PERSELLO , RONNY HÄNSCH , GEMINE VIVONE ,
KAIQIANG CHEN , ZHIYUAN YAN , DEKE TANG, HAI HUANG ,
MICHAEL SCHMITT, AND XIAN SUN
2023 IEEE GRSS Data Fusion Contest:
Large-Scale Fine-Grained Building Classification
for Semantic Urban Reconstruction
B
uildings are essential components
of urban areas. While
research on the extraction and 3D
reconstruction of buildings is wide -
ly conducted, information on the
fine-grained roof types of buildings
is usually ignored. This limits
the potential of further analysis,
e.g., in the context of urban planning
applications. The fine-grained
classification of building roof types
from satellite images is a highly
challenging task due to ambiguous
visual features within the satellite
imagery. The lack of corresponding fine-grained building
classification datasets further increases the difficulty.
The 2023 IEEE Geoscience and Remote Sensing Society
(GRSS) Data Fusion Contest (DFC23) (see Figure 1),
organized by the Image Analysis and Data Fusion Technical
Committee (IADF TC) of the GRSS, the Aerospace Information
Research Institute under the Chinese Academy
of Sciences, the Universität der Bundeswehr München,
and GEOVIS Earth Technology Co., Ltd., aims to push
current research on building extraction, classification,
and 3D reconstruction toward urban reconstruction with
fine-grained semantic information of roof types and
height information.
To this aim, the DFC23 establishes a large-scale, finegrained,
and multimodal benchmark for the classification
of building roof types. It consists of two challenging
competition tracks investigating the fusion of optical
and synthetic aperture radar (SAR) data while focusing
on roof type classification and building height estimation,
respectively. The data provided by the DFC23 includes
several novel properties.
Digital Object Identifier 10.1109/MGRS.2023.3240233
Date of current version: 31 March 2023
94
◗ Globally distributed and large scale:
A novel large-scale urban building
classification and reconstruction
dataset is provided. Buildings
are distributed across 17
cities on six continents to cover
a wide range of different building
styles. This allows capturing
the heterogeneity of cities in different
continents with various
landforms.
◗ Fine-grained roof type categories:
The buildings are labeled according
to a detailed (fine-grained)
categorization of roof types. The DFC23 provides
nearly 300,000 instances with 12 different types of
building roofs, rendering building classification significantly
more challenging.
◗ Multimodal data: To facilitate multimodal data fusion,
not only optical imagery but also SAR images
are provided. The information captured by these different
modalities can be jointly exploited, potentially
resulting in the development of more accurate building
extraction and classification models.
The contest is designed as a benchmark competition
following the previous editions [1], [2], [3], [4],
[5], [6], [7] and will consist of the following two parallel
tracks:
1) " Track 1: Building Detection and Roof Type Classification "
2)
" Track 2: Multi-Task Learning of Joint Building Extraction
and Height Estimation. "
Track 1: Building Detection and Roof Type Classification:
This track focuses on the detection and classification
of building roof types from high-resolution
optical satellite imagery and SAR images. The SAR
and optical modalities are expected to provide complementary
information. The given dataset covers
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE MARCH 2023

https://orcid.org/0000-0002-2936-6765 https://orcid.org/0000-0001-9542-0638 https://orcid.org/0000-0003-3742-5398 https://orcid.org/0000-0002-8314-2375 https://orcid.org/0000-0002-7114-6758 https://orcid.org/0000-0001-8745-8142 https://orcid.org/0000-0002-0038-9816

IEEE Geoscience and Remote Sensing Magazine - March 2023

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2023