Dynamic mode decomposition (DMD) is a widely used technique for analyzing flow dynamics, particularly with experimental data. Traditionally, in experimental applications, the method frequently relies on time-resolved particle image velocimetry (PIV) data to construct an embedding to track incremental changes in the flow state. In this study, we propose an extension of DMD to enable the extraction of DMD modes from standard (non-time-resolved) PIV data. Standard PIV data typically consists of paired snapshots of particle positions taken with a small time delay, used to compute instantaneous velocity fields. However, the time delay between successive pairs is often too large to effectively capture the evolution of the velocity field. To address this limitation, we associate particle-position data with flow dynamics by introducing a gap-based DMD (gDMD) formulation. In this approach, the local distribution of seed particles is mapped to directional gap fields, which are then used as observables for extracting velocity-associated DMD modes. The method is showcased using two direct numerical simulation databases: a confined two-dimensional cylinder wake and the fluidic pinball configuration, together with PIV data of a rectangular jet. In the numerical case, artificial PIV data were generated by seeding the flow with passive tracers. Our results indicate that dominant DMD modes can be accurately predicted even with significant time delays between successive snapshot pairs—an interval that would typically hinder the application of standard DMD. This approach broadens the applicability of DMD, providing a robust tool for analyzing flow dynamics with non-time-resolved PIV data.