Urban Change Detection for Multispectral Earth Observation Using Convolutional Neural Networks

By Rodrigo Caye Daudt, Bertrand Le Saux, Alexandre Boulch, Yann Gousseau. IEEE International Geoscience and Remote Sensing Symposium, 2018

452 citations recorded by the source corpus.

The Copernicus Sentinel-2 program now provides multispectral images at a global scale with a high revisit rate. In this paper we explore the usage of convolutional neural networks for urban change detection using such multispectral images. We first present the new change detection dataset that was used for training the proposed networks, which will be openly available to serve as a benchmark. The Onera Satellite Change Detection (OSCD) dataset is composed of pairs of multispectral aerial images, and the changes were manually annotated at pixel level. We then propose two architectures to detect changes, Siamese and Early Fusion, and compare the impact of using different numbers of spectral channels as inputs. These architectures are trained from scratch using the provided dataset.

Subfields: Atmospheric Science, Ecology, Media Technology

View Paper: https://www.semanticscholar.org/paper/18755de13d2d868255bbf332fa80ec218163c74e

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 Ecology

In Media Technology

Citations in the corpus

Optical Remote Sensing Image Change Detection Based on Attention Mechanism and Image Difference (2021, methodology): “The proposed approach was evaluated on ﬁve change detection data sets: the season-varying data set [48], the WHU building change detection data set [1], the SZTAKI AirChange benchmark set [49], the Onera Satellite change detection (OSCD) data set [50], and the LEVIR-CD data set [51].”
DynamicEarthNet: Daily Multi-Spectral Satellite Dataset for Semantic Change Segmentation (2022, background): “Most relevant datasets focus on binary change detection which is agnostic to specific types of change [3,13].”
FloodNet: A High Resolution Aerial Imagery Dataset for Post Flood Scene Understanding (2020, methodology): “Onera Satellite Change Detection (OSCD) dataset was proposed in [17] which consists of multispectral aerial images to detect urban growth and changes with time.”
Self-Supervised Change Detection in Multiview Remote Sensing Images (2021, methodology): “1) OSCD S2S2/ S1S1/ S1S2/ L8S2: The Onera Satellite Change Detection (OSCD) dataset [45] was created for bitemporal change detection using Sentinel-2 images acquired between 2015 and 2018.”
Multitask learning for large-scale semantic change detection (2018, methodology): “Using the CNN architecture labelled EF by Daudt et al. (2018b), we have achieved excellent numeric results which discouraged the usage of more complex methods which would lead to even more extreme overﬁtting.”
Towards Geospatial Foundation Models via Continual Pretraining (2023, methodology): “We conduct experiments on both the Onera Satellite Change Detection (OSCD [5]) in Table 3 and DSIFN [43] in Table 4.”
Self-Supervised Representation Learning for Remote Sensing Image Change Detection Based on Temporal Prediction (2020, methodology): “org/) provided by the Copernicus Sentinel-2 program [69] [see Figure A5a,b].”
A Deeply Supervised Attention Metric-Based Network and an Open Aerial Image Dataset for Remote Sensing Change Detection (2021, background): “Although DL-based models have shown strong feature extraction ability and achieved great process in CD applications, some remarkable CD approaches based on the DL framework have been developed [6], [21]–[23].”
Deep Learning-Based Change Detection in Remote Sensing Images: A Review (2022, methodology): “The datasets that fall under it are Southwest U. S. Change Detection Dataset, MtS-WH, Taizhou dataset [73], Onera Satellite Change Detection dataset [74], and NASA Earth Observatory Change Datasets.”
SpectralGPT: Spectral Remote Sensing Foundation Model (2023, methodology): “For the change detection task, we utilize the OSCD dataset [43].”
SwinSUNet: Pure Transformer Network for Remote Sensing Image Change Detection (2022, methodology): “Consistent with [46], we used 14 pairs of images for training and the rest for testing.”
End-to-End Change Detection for High Resolution Satellite Images Using Improved UNet++ (2019, methodology): “Rodrigo et al. [44] ﬁrstly proposed two end-to-end PB-DLCD architectures, where the CNNs could be trained from scratch using only the provided datasets.”
A Spatial-Temporal Attention-Based Method and a New Dataset for Remote Sensing Image Change Detection (2020, methodology): “OSCD, introduced last year, has also driven a few studies [67,69].”
Fully Convolutional Siamese Networks for Change Detection (2018, methodology): “Nevertheless, there are available pixelwise annotated change detection datasets available that can be used to train supervised machine learning systems that detect changes in image pairs, such as the Onera Satellite Change Detection dataset 1 presented in [3] and the Air Change dataset [4].”
Transformers in Remote Sensing: A Survey (2022, methodology): “SwinSUNet [56] Change Detection CCD [82], WHU [60], OSCD [98], HRSCD [99] F1 score Introduces a Swin transformer-based network with a Siamese U-shaped structure having encoder, fusion and decoder modules.”
Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Data (2021, methodology): “The Onera Satellite Change Detection (OSCD) dataset [8] is composed of 24 pairs of multispectral images from Sentinel-2.”