CloudSEN12 - a global dataset for semantic understanding of cloud and cloud shadow in Sentinel-2

Aybar, César; Ysuhuaylas, Luis; Loja, Jhomira; Gonzales, Karen; Herrera, Fernando; Yali, Roy; Flores, Angie; Diaz, Lissette; Cuenca, Nicole; Espinoza, Wendy; Prudencio, Fernando; Llactayo, Valeria; Loaiza, David Montero; Sudmanns, Martin; Tiede, Dirk; Mateo-García, Gonzalo; Gómez-Chova, L.

doi:10.31223/x5s35g

Cited by 2 publications

(2 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The other published methods are those provided by Aybar et al [13] and are not re-computed, with the exception of the UN-etMobV2 model, for which the authors' github package [49] was used. Instead of recomputing masks for each method, the masks provided alongside the CloudSEN12 dataset are used.…”

Section: A Model Improvement Experimentsmentioning

confidence: 99%

Sensor Independent Cloud and Shadow Masking With Partial Labels and Multimodal Inputs

Francis

2024

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

A paradigm shift is underway in Earth Observation, as deep learning replaces other methods for many predictive tasks. Nevertheless, most deep learning classification models for Earth Observation are limited by their specificity with respect to both the sensors used (inputs) and classes predicted (outputs), leading to models which only perform well for specific satellites and on specific datasets. Cloud masking is typical of this, but is one of the most important tasks to generalise across sensors, given that it is required for all optical instruments. This work sets out a framework to relax deep learning's constraints on specific inputs and outputs, using cloud and shadow masking as a case-study. Centrally, a model which is sensor independent, and which can simultaneously learn from different labelling schemes is developed. The model, Spectral ENcoder for SEnsor Independence version 2 (SEnSeI-v2) extends the original version, by permitting multimodal data (in this case Sentinel-1 SAR imagery and a DEM) to be ingested, along with several other architectural improvements. SEnSeI-v2, attached to SegFormer, is shown to have state-of-the-art performance, whilst being usable on a range of multispectral band combinations, alongside SAR and DEM inputs, without retraining. The labelling schemes of eight datasets are not made compatible through a reductive approach (e.g. converting to cloud vs. non-cloud), rather, an ambiguous cross-entropy loss is introduced that allows the model to learn from the different labelling schemes without sacrificing the class distinctions of each, leading to a model which predicts all of the constituent classes of the different datasets.

show abstract

Section: A Model Improvement Experimentsmentioning

confidence: 99%

Sensor Independent Cloud and Shadow Masking With Partial Labels and Multimodal Inputs

Francis

2024

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

show abstract

“…on the data set of [35]. Moreover, recent publications have provided novel large-scale data sets for cloud detection or removal in time-series [12], [13], [36], which may serve for an extended version of our analysis. With respect to the cloud detector algorithm, s2cloudless was chosen for being commonly deployed, easily applicable and performing well [37], [38].…”

Section: ) Outliers As Distractorsmentioning

confidence: 99%

Explaining the Effects of Clouds on Remote Sensing Scene Classification

Gawlikowski

Ebel

Schmitt

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Most of Earth is covered by haze or clouds, impeding the constant monitoring of our planet. Preceding works have documented the detrimental effects of cloud coverage on remote sensing applications and proposed ways to approach this issue. However, up-to-now, little effort has been spent on understanding how exactly atmospheric disturbances impede the application of modern machine learning methods to Earth observation data. Specifically, we consider the effects of haze and cloud coverage on a scene classification task. We provide a thorough investigation of how classifiers trained on cloud-free data fail once they encounter noisy imagery -a common scenario encountered when deploying pretrained models for remote sensing to real use cases. We show how and why remote sensing scene classification suffers from cloud coverage. Based on a multi-stage analysis, including explainability approaches applied to the predictions, we work out four different types of effects that clouds have on the scene prediction. The contribution of our work is to deepen the understanding of the effects of clouds on common remote sensing applications, and consequently guide the development of more robust methods.

show abstract

CloudSEN12 - a global dataset for semantic understanding of cloud and cloud shadow in Sentinel-2

Cited by 2 publications

References 53 publications

Sensor Independent Cloud and Shadow Masking With Partial Labels and Multimodal Inputs

Sensor Independent Cloud and Shadow Masking With Partial Labels and Multimodal Inputs

Explaining the Effects of Clouds on Remote Sensing Scene Classification

Contact Info

Product

Resources

About