Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a novel unsupervised change detection method using Remote Sensing Foundation Models (RSFMs) that leverages a ResNet architecture combined with a Feature Pyramid Network (FPN) to identify anomalies in satellite imagery. By detecting subtle semantic shifts in the latent space between successive orbital passes, the method operates without the need for extensive training or bespoke models, thus optimizing onboard resources for disaster monitoring. The results demonstrate that this approach can achieve high-fidelity feature extraction and efficient image-level generation, significantly enhancing the capabilities of satellite missions in real-time anomaly detection.
Unsupervised change detection using RSFMs can revolutionize disaster monitoring by eliminating the need for expensive labeled datasets and bespoke model training.
Remote Sensing Foundation Models (RSFMs) have emerged as a powerful alternative to supervised models for Earth Observation, allowing satellites to autonomously trigger high-resolution captures or adjust tasking parameters upon detecting an anomaly, thereby maximizing the utility of the mission's limited power and computational resources. RSFMs are versatile, unified encoders that optimize onboard storage for multiple orbital applications while ensuring high-fidelity feature extraction. In particular, unsupervised change detection with RSFMs offers a well-informed and transformative path for disaster monitoring without expensive labels. In this paper, we present a novel unsupervised detection method based on ResNet (RSFM) + FPN which identifies a wide spectrum of anomalies by detecting subtle semantic shifts in the latent space between successive orbital passes. By relying on an untrained FPN architecture and its intrinsic priors, the system achieves efficient image-level generation and higher resolution mapping with minimal effort (training-free) compared to previous proposals (patch-based, trained). And by replacing tailored models with RSFMs, we can achieve comparable results through an approach that eliminates the need for bespoke training and extensive development effort and adds customization, while ensuring high-performance generalization across diverse terrains and sensors.