Algorithm Selection (AS) is a research area that aims to automatically select the best algorithm from a portfolio for a given optimization problem. While most contemporary AS methods rely on feature-based algorithm selection, recent studies have focused on landscape feature analysis. To address this limitation, we propose a new AS system that uses images of problem function visualization as input data and eliminates the need for feature analysis. We evaluated our model’s performance against the Single Best Solver (SBS) and existing traditional AS method, but collecting image data posed some challenges. Nevertheless, our study results were competitive and promising. Our research introduces a novel approach to AS for black-box optimization and explores potential avenues for improving the model in future work.