The recognition and verification of marine electrical connector types constitute a critical process in ship electrical outfitting and electrical equipment installation, as their accuracy directly impacts the reliability of marine electrical systems. To address the issues of missing pins and reflective interference under low-light outfitting conditions in existing methods that rely on local pinhole feature recognition, this study proposes a method based on Maximally Stable Extremal Regions (MSER) and triangulation to construct topological feature vectors, combined with a lightweight SVM classification algorithm for connector type recognition. Cross-validation experiments demonstrate that the use of topological feature vectors achieves a higher recognition rate for marine electrical connector types compared to local pinhole feature-based methods.