Bigearthnet: A Large-Scale Benchmark Archive for Remote Sensing Image Understanding

By Gencer Sumbul, Marcela Charfuelan, Begüm Demir, Volker Markl. 2019

594 citations recorded by the source corpus.

This paper presents the BigEarthNet that is a new large-scale multi-label Sentinel-2 benchmark archive. The BigEarthNet consists of 590, 326 Sentinel-2 image patches, each of which is a section of i) 120 × 120 pixels for 10m bands; ii) 60×60 pixels for 20m bands; and iii) 20×20 pixels for 60m bands. Unlike most of the existing archives, each image patch is annotated by multiple land-cover classes (i.e., multi-labels) that are provided from the CORINE Land Cover database of the year 2018 (CLC 2018). The BigEarthNet is significantly larger than the existing archives in remote sensing (RS) and thus is much more convenient to be used as a training source in the context of deep learning. This paper first addresses the limitations of the existing archives and then describes the properties of the BigEarthNet. Experimental results obtained in the framework of RS image scene classification problems show that a shallow Convolutional Neural Network (CNN) architecture trained on the BigEarthNet provides much higher accuracy compared to a state-of-the-art CNN model pre-trained on the ImageNet (which is a very popular large-scale benchmark archive in computer vision). The BigEarthNet opens up promising directions to advance operational RS applications and research in massive Sentinel-2 image archives.

Subfields: Atmospheric Science, Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

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

In Atmospheric Science

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Continual Barlow Twins: Continual Self-Supervised Learning for Remote Sensing Semantic Segmentation (2022, background): “[18] learns visual representations inferring information on the visible spectrum from the other bands on BigEarthNet [19].”
The color out of space: learning self-supervised representations for Earth Observation imagery (2020, methodology): “III-C on BigEarthNet.”
SatMAE: Pre-training Transformers for Temporal and Multi-Spectral Satellite Imagery (2022, methodology): “While several large-scale satellite image datasets have been carefully curated in the past few years, including Functional Map of the World (fMoW) [17], BigEarthNet [18], xView [19], SpaceNet [20], annotating these datasets requires specialized skills and is more expensive than traditional computer ”
An Empirical Study of Remote Sensing Pretraining (2022, background): “It contains 100,0848 nonoverlapping scenes, exceeding the competitive counterpart fMoW [54] and BigEarthNet [55], which separately includes 132,716 and 590,326 scenes.”
Self-Supervised Learning in Remote Sensing: A review (2022, methodology): “4) Transfer Learning: In addition, we report transfer learning performance by linear classiﬁcation and ﬁne-tuning on EuroSat with a ResNet18 pre-trained on BigEarthNet.”
SEN12MS - A Curated Dataset of Georeferenced Multi-Spectral Sentinel-1/2 Imagery for Deep Learning and Data Fusion (2019, background): “BigEarthNet 590,326 up to 120× 120 Sentinel-2 43 Corine Land Cover classes over Europe (Sumbul et al., 2019)”
Spotting Insects from Satellites: Modeling the Presence of Culicoides Imicola Through Deep CNNs (2019, background): “In this regards - further than gathering more data - we envision a network pre-training on crop and land classification directly, for which large datasets [27] are available currently.”
Towards Geospatial Foundation Models via Continual Pretraining (2023, methodology): “Notably, we find that our model maintains strong performance on BigEarthNet, even when only given 1% of the training data.”
Self-Supervised Learning of Remote Sensing Scene Representations Using Contrastive Multiview Coding (2021, methodology): “BigEarthNet is a dataset of multispectral satellite images for multi-label land use classification.”
Satlas: A Large-Scale, Multi-Task Dataset for Remote Sensing Image Understanding (2022, methodology): “Many of these focus on scene and patch classi-ﬁcation tasks: the UC Merced Land Use (UCM) [53] and BigEarthNet [48] datasets involve land cover classiﬁcation with 21 and 43 categories respectively, while the AID [52], Million-AID [37], RESISC45 [18], and Functional Map of the World (FMoW) [21]…”
SpectralGPT: Spectral Remote Sensing Foundation Model (2023, methodology): “For the multi-label RS scene classification task, we utilize the BigEarthNet-S2 dataset [35].”
Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Data (2021, methodology): “To the best of our knowledge, this is the ﬁrst time an unsupervised method obtains higher accuracy than Ima-geNet pre-training on BigEarthNet with 100% of the labels.”
In-domain representation learning for remote sensing (2019, background): “BigEarthNet (Sumbul et al., 2019) is a challenging large-scale multi-spectral dataset consisting of 590,326 image patches from the Sentinel-2 satellite.”
Self-Supervised Remote Sensing Feature Learning: Learning Paradigms, Challenges, and Future Works (2022, methodology): “The literature [70] assessed the transferability of self-supervised learned features by the EuroSAT [250] and BigEarthNet [65] datasets using the linear probe and fine-tuning.”
Object Detection and Image Segmentation with Deep Learning on Earth Observation Data: A Review-Part I: Evolution and Recent Trends (2020, methodology): “NWPU RESISC45 [155] IR LULC multiple platforms optical high VGG-16 [155] EuroSAT [156] IR LULC Sentinel 2 multispectral medium Inception-V1 and ResNet-50 [156] BigEarthNet [116,117] IR LULC Sentinel 2 multispectral medium ResNet-50 [117] So2Sat LCZ42 [157] IR local climate zones Sentinel 1+2 mltspec”
Remote Sensing Image Scene Classification Meets Deep Learning: Challenges, Methods, Benchmarks, and Opportunities (2020, background): “images [49], [74], [79], [80], [85], [124]–[131] to facilitate this field forward.”