OceanPile: Large-Scale Multimodal Corpus for Ocean AI

Researchers have introduced OceanPile, a large-scale multimodal corpus designed to overcome data fragmentation in ocean science. Ocean data are scattered across sources, noisy, and weakly labeled, hindering AI applications. OceanPile includes OceanCorpus, which integrates sonar data, underwater imagery, and marine science visualizations into a unified schema. This dataset aims to enable Multimodal Large Language Models (MLLMs) to tackle ocean-related tasks, such as climate modeling and biodiversity monitoring. The work addresses a critical bottleneck in applying AI to marine environments.