A Novel Benchmark Framework for Neural Embeddings in Earth Observation

ICLR 2026 Conference Submission17975 Authors

Keywords: neural embeddings, benchmark framework, spatio-temporal data, Earth observation

TL;DR: a novel benchmarking pipeline for spatio-temporal neural embeddings

Abstract: We introduce a novel benchmark framework for evaluating (lossy) neural compression and representation learning in the context of Earth Observation (EO). Our approach builds on fixed-size embeddings that act as compact, task-agnostic representations applicable to a broad range of downstream tasks. Our benchmark comprises three core components: (i) an evaluation pipeline built around reusable embeddings, (ii) a new challenge mode with a hidden-task leaderboard designed to mitigate pretraining bias, and (iii) a scoring system that balances accuracy and stability. To support reproducibility, we release a curated multispectral, multitemporal EO dataset. We present initial results from a public challenge at a workshop and conduct ablations with state-of-the-art foundation models. Our benchmark provides a first step towards community-driven, standardized evaluation of neural embeddings for EO and beyond.

Primary Area: datasets and benchmarks

Submission Number: 17975