<p>In the context of increasing frequency and impact of flood events, traditional methods for estimating flood depth have become insufficient to meet current demands, leading to a gradual shift toward machine learning approaches. This article reviews, for the first time, the applications of machine learning models—including both single and hybrid models—in flood depth estimation, referencing 108 relevant studies. Through statistical analysis, this research explored the most commonly used machine learning models and their primary data sources. Building on this foundation, we also examined the potential for integrating machine learning methods with smart city frameworks and artificial intelligence large models for flood depth calculations. The findings indicate that machine learning models excel in handling large-scale complex data and nonlinear relationships, and their performance can be further optimized through combinations with various models, significantly enhancing the accuracy and efficiency of flood depth estimation. However, these models also face challenges such as data dependency, model interpretability, and transferability. This review reveals the potential of applying machine learning models in flood depth estimation, providing directions for future research and reliable support for disaster prevention and reduction efforts.</p>