WaveMAE: Wavelet decomposition Masked Auto-Encoder for Remote Sensing

TL;DR

WaveMAE introduces a wavelet-based masked autoencoder for remote sensing that effectively captures scale-aware features and geospatial priors, leading to improved performance across various downstream tasks.

Contribution

The paper proposes WaveMAE, a novel wavelet-based masked autoencoder with geo-conditioned positional encoding for enhanced remote sensing imagery analysis.

Findings

WaveMAE outperforms previous methods on segmentation and regression tasks.

A lightweight WaveMAE variant achieves state-of-the-art results with only 26.4% of parameters.

Pretraining on the same dataset ensures fair comparison and robust evaluation.

Abstract

Self-supervised learning (SSL) has recently emerged as a key strategy for building foundation models in remote sensing, where the scarcity of annotated data limits the applicability of fully supervised approaches. In this work, we introduce WaveMAE, a masked autoencoding framework tailored for multispectral satellite imagery. Unlike conventional pixel-based reconstruction, WaveMAE leverages a multi-level Discrete Wavelet Transform (DWT) to disentangle frequency components and guide the encoder toward learning scale-aware high-frequency representations. We further propose a Geo-conditioned Positional Encoding (GPE), which incorporates geographical priors via Spherical Harmonics, encouraging embeddings that respect both semantic and geospatial structure. To ensure fairness in evaluation, all methods are pretrained on the same dataset (fMoW-S2) and systematically evaluated on the diverse downstream tasks of the PANGAEA benchmark, spanning semantic segmentation, regression, change detection, and multilabel classification. Extensive experiments demonstrate that WaveMAE achieves consistent improvements over prior state-of-the-art approaches, with substantial gains on segmentation and regression benchmarks. The effectiveness of WaveMAE pretraining is further demonstrated by showing that even a lightweight variant, containing only 26.4% of the parameters, achieves state-of-the-art performance. Our results establish WaveMAE as a strong and geographically informed foundation model for multispectral remote sensing imagery.