Land cover mapping at very high resolution with rotation equivariant CNNs: towards small yet accurate models

By Diego Marcos, Michele Volpi, Benjamin Kellenberger, Devis Tuia. 2018

261 citations recorded by the source corpus.

Abstract In remote sensing images, the absolute orientation of objects is arbitrary. Depending on an object’s orientation and on a sensor’s flight path, objects of the same semantic class can be observed in different orientations in the same image. Equivariance to rotation, in this context understood as responding with a rotated semantic label map when subject to a rotation of the input image, is therefore a very desirable feature, in particular for high capacity models, such as Convolutional Neural Networks (CNNs). If rotation equivariance is encoded in the network, the model is confronted with a simpler task and does not need to learn specific (and redundant) weights to address rotated versions of the same object class. In this work we propose a CNN architecture called Rotation Equivariant Vector Field Network (RotEqNet) to encode rotation equivariance in the network itself. By using rotating convolutions as building blocks and passing only the values corresponding to the maximally activating orientation throughout the network in the form of orientation encoding vector fields, RotEqNet treats rotated versions of the same object with the same filter bank and therefore achieves state-of-the-art performances even when using very small architectures trained from scratch. We test RotEqNet in two challenging sub-decimeter resolution semantic labeling problems, and show that we can perform better than a standard CNN while requiring one order of magnitude less parameters.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

General E(2)-Equivariant Steerable CNNs (2019, neutral): “04 17 regular/vector ψ1 conv −−→ ρreg vector pool −−−−−−→ ψ1 vector field [13, 37] 2.”
Swin Transformer Embedding UNet for Remote Sensing Image Semantic Segmentation (2022, unknown): “Following [38] and [66]–[68], we select 11 images for training (image IDs: 1, 3, 5, 7, 13, 17, 21, 23, 26, 32, and 37), ﬁveimages for testing (image IDs: 11, 15, 28, 30, and 34), and crop them to 256 × 256, respectively.”
UNetFormer: A UNet-like transformer for efficient semantic segmentation of remote sensing urban scene imagery (2021, unknown): “However, within fine-resolution remotely sensed urban scene images, complicated patterns and human-made objects occur frequently (Kampffmeyer et al., 2016; Marcos et al., 2018) (Audebert et al., 2018).”
iSAID: A Large-scale Dataset for Instance Segmentation in Aerial Images (2019, background): “There are numerous lucrative application areas of Earth Vision research, including security and surveillance [28], urban planning [22], precision agriculture, land type classi-ﬁcation [2] and change detection [15].”
LoveDA: A Remote Sensing Land-Cover Dataset for Domain Adaptive Semantic Segmentation (2021, methodology): “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]”