This study uses a distributed accelerometer network to detect and localize vehicles on a bridge structure. Two methods were used in this research: probabilistic frequency domain analysis and wavelet scattering network. The accuracy of the detection methods was evaluated against user-labeled data using camera-captured images. The proposed methodology is validated for a bridge on a Trans-Canada Highway in the Province of New Brunswick. The case-study bridge is instrumented using a permanent long-term monitoring system, which includes accelerometers, strain gauges, and cameras. The presented method is valuable in identifying vehicles on a bridge structure for traffic monitoring and, in general, developing a bridge-traffic digital twin. The proposed method is also integral to Bridge Weigh-in-Motion (BWIM) systems. The methods introduced in this paper are particularly valuable for implementing BWIM systems on bridges with multiple vehicle presence.