Toward Foundation Models for Earth Monitoring: Proposal for a Climate Change Benchmark

Recent progress in self-supervision shows that pre-training large neural\nnetworks on vast amounts of unsupervised data can lead to impressive increases\nin generalisation for downstream tasks. Such models, recently coined as\nfoundation models, have been transformational to the field of natural language\nprocessing. While similar models have also been trained on large corpuses of\nimages, they are not well suited for remote sensing data. To stimulate the\ndevelopment of foundation models for Earth monitoring, we propose to develop a\nnew benchmark comprised of a variety of downstream tasks related to climate\nchange. We believe that this can lead to substantial improvements in many\nexisting applications and facilitate the development of new applications. This\nproposal is also a call for collaboration with the aim of developing a better\nevaluation process to mitigate potential downsides of foundation models for\nEarth monitoring.\n