Recreational boats contribute to underwater noise pollution in shallow coastal regions, potentially affecting marine ecosystems. Reliable, automated methods for detecting vessel passages are essential for continuous acoustic monitoring and environmental assessment. This study presents the development and evaluation of an energy detector (ED) based on the short-time-average/long-time-average (STA/LTA) method, originally from seismology, adapted to detect transient acoustic events caused by small vessels. Broadband and multiband variants of the detector were designed and tested on real underwater recordings from the Swedish west coast, collected via hydrophones in shallow water. Performance was assessed across a range of STA and LTA window durations, frequency bands, and detection thresholds. The receiver operating characteristic (ROC) curve and area under the curve (AUC) served as key evaluation metrics. Results demonstrated that good detection performance was achieved using an STA of 3 min and an LTA spanning the entire recording length. The multiband detector slightly outperformed the broadband version, especially at higher false positive rates, indicating improved spectral discrimination. These findings suggest that the STA/LTA method offers a practical and effective approach for long-term, automated vessel monitoring in dynamic marine acoustic environments.