Abstract: Ensuring fairness in computer vision models is critical for their ethical deployment in real-world applications. However, these models often learn spurious correlations between sensitive ...