<p>Wave breaking type is a fundamental indicator of nearshore hydrodynamic processes, directly reflecting wave energy dissipation mechanisms. With the advancement of shore-based video monitoring, remote sensing has emerged as an efficient tool for identifying wave breaking types. However, existing studies predominantly rely on static single-frame imagery, limiting the ability to capture the dynamic evolution of breaking events. In this work, we present the first publicly available video dataset dedicated to wave breaking type classification. The dataset comprises 9,000 labeled wave breaking clips collected from 15 cameras across six morphologically diverse coastal sites, encompassing three primary breaking types: Spilling, Plunging, and Surging. To enhance the dataset’s quality and consistency, a rigorous data curation workflow was implemented, including video segmentation, cropping, labeling, and frame extraction. Classification experiments using a well-established deep learning architecture combing CNN and RNN achieved state-of-the-art performance.</p>