DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Images

By Ilke Demir, Krzysztof Koperski, David Lindenbaum, Guan Pang, Jing Huang, Saikat Basu, Forest Hughes, Devis Tuia, Ramesh Raskar. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2018

1,111 citations recorded by the source corpus.

We present the DeepGlobe 2018 Satellite Image Understanding Challenge, which includes three public competitions for segmentation, detection, and classification tasks on satellite images (Figure 1). Similar to other challenges in computer vision domain such as DAVIS[21] and COCO[33], DeepGlobe proposes three datasets and corresponding evaluation methodologies, coherently bundled in three competitions with a dedicated workshop co-located with CVPR 2018. We observed that satellite imagery is a rich and structured source of information, yet it is less investigated than everyday images by computer vision researchers. However, bridging modern computer vision with remote sensing data analysis could have critical impact to the way we understand our environment and lead to major breakthroughs in global urban planning or climate change research. Keeping such bridging objective in mind, DeepGlobe aims to bring together researchers from different domains to raise awareness of remote sensing in the computer vision community and vice-versa. We aim to improve and evaluate state-of-the-art satellite image understanding approaches, which can hopefully serve as reference benchmarks for future research in the same topic. In this paper, we analyze characteristics of each dataset, define the evaluation criteria of the competitions, and provide baselines for each task.

Subfields: Computer Vision and Pattern Recognition, Media Technology, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/15a148957469bc8b91bd7cc31aa1f0c6584a1571

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

In Media Technology

In Ocean Engineering

Citations in the corpus

D-LinkNet: LinkNet with Pretrained Encoder and Dilated Convolution for High Resolution Satellite Imagery Road Extraction (2018, methodology): “In the DeepGlobe Road Extraction Challenge, we found that transfer learning can accelerate our network convergence and make it have better performance with almost no extra cost.”
Understanding Deep Learning Techniques for Image Segmentation (2019, neutral): “Natural Berkeley Segmentation Dataset [140] Scenes PASCAL VOC [54] Stanford Background Dataset [72] Microsoft COCO [122] MIT Scene parsing data(ADE20K) [222, 223] Semantic Boundaries Dataset [75] Microsoft Research Cambridge Object Recognition Image Database (MSRC) [188] Video Densely Annotated Vide”
TernausNetV2: Fully Convolutional Network for Instance Segmentation (2018, methodology): “To evaluate our model performance the predicted masks for the test images should be upload into DeepGlobe website [3, 9].”
Continual Learning With Structured Inheritance for Semantic Segmentation in Aerial Imagery (2021, methodology): “2) Ablation Experiments: To evaluate the effectiveness of our method, we conduct comprehensive ablation experiments for continual learning tasks starting DeepGlobe [41] to Potsdam [40].”
LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentation (2021, background): “The existing HSR land-cover datasets such as the Gaofen Image Dataset (GID) [38], DeepGlobe [9], Zeebruges [24], and Zurich Summer [43] contain large-scale images with pixel-wise annotations, thus promoting the development of fully convolutional networks (FCNs) in the ﬁeld of remote sensing [46, 49]”
Global and Local Contrastive Self-Supervised Learning for Semantic Segmentation of HR Remote Sensing Images (2021, background): “However, surprisingly, Ours_noGlobal achieves the worst results on the DGLC dataset, which indicates that the global module is extremely important on this dataset.”
Feature Pyramid Network for Multi-class Land Segmentation (2018, methodology): “These images have 50cm pixel resolution, collected by DigitalGlobe’s satellite [2, 5].”
Self-Supervised Remote Sensing Feature Learning: Learning Paradigms, Challenges, and Future Works (2022, background): “(DGLCC) [255], a land cover classification dataset containing 1146 images annotated in seven classes.”
SEN12MS - A Curated Dataset of Georeferenced Multi-Spectral Sentinel-1/2 Imagery for Deep Learning and Data Fusion (2019, background): “Therefore, massive databases of labeled imagery such as ImageNet (Deng et al., 2009) do not yet exist in the remote sensing domain, although there have been ﬁrst steps into that direction; a certainly non-exhaustive overview of existing scientiﬁc datasets of annotated remote sensing imagery can be…”
Semantic Stereo for Incidental Satellite Images (2018, background): “Several recent public challenge datasets focusing on satellite images have been instrumental in benchmarking and advancing the state of the art for semantic labeling tasks [5, 6, 7, 8] and multiview 3D reconstruction [9, 10].”
FALSE: False Negative Samples Aware Contrastive Learning for Semantic Segmentation of High-Resolution Remote Sensing Image (2022, positive): “The experiments were selected from the public RSI semantic segmentation dataset ISPRS Potsdam [24], competition dataset CVPR DGLC [25], and Xiangtan dataset [10] from the Gaofen-2 satellite covering Xiangtan, China.”
Satlas: A Large-Scale, Multi-Task Dataset for Remote Sensing Image Understanding (2022, methodology): “…existing remote sensing datasets for object detection, instance segmentation, and semantic segmentation like DOTA [51], iSAID [54], and Deep-Globe [23] contain less than 10K images each, compared to the 328K images in COCO and millions used to train CLIP; the small size of these datasets means we…”
FAIR1M: A Benchmark Dataset for Fine-grained Object Recognition in High-Resolution Remote Sensing Imagery (2021, methodology): “There are also datasets dedicated to building detection, such as the Semcity toulouse dataset [43], ISPRS benchmark on urban object detection and 3D building reconstruction dataset [44], Inria Aerial Image Labeling dataset [36] and DeepGlobe 2018 dataset [9].”
UAVid: A semantic segmentation dataset for UAV imagery (2018, methodology): “The state-of-the-art DeepGlobe Land Cover Classification dataset [14] has 1146 satellite images for rural areas of size 2448 × 2448, which is about 2.”
xBD: A Dataset for Assessing Building Damage from Satellite Imagery (2019, background): “Relatively few datasets [3, 11] provide information of this granularity.”
Multitask learning for large-scale semantic change detection (2018, methodology): “The amount of labelled pixels and surface area for land cover classification is also about 8 times larger in the proposed HRSCD dataset than in the DeepGlobe Land Cover Classification dataset (Demir et al., 2018), both of the datasets containing RGB images of the same spatial resolution (50 cm”
RSVQA: Visual Question Answering for Remote Sensing Data (2020, background): “This architecture has been successfully used in a wide range of work in the remote sensing community [8], [17], [33].”