This study evaluates the use of automated video-based classification to analyze traffic flows and travel patterns at four major intersections in Maribor, Slovenia. Using the DataFromSky software, traffic data were collected during the morning peak period on a working day in April 2025. The study focused on identifying the modal split and micro-distribution of movements as part of the traditional four-step transportation planning model. Five categories of road users were analyzed: cars, buses, motorcycles, cyclists, and pedestrians. Results revealed a high share of car traffic at all sites, while the city center intersection (Site 4) showed a significantly higher share of cyclists and pedestrians, reflecting the influence of pedestrian zones and cycling infrastructure. Origin–destination matrices highlighted the concentration of travel toward the city center and emphasized key cycling corridors. The accuracy of vehicle type classification exceeded 99%, confirming the software’s reliability for traffic monitoring. The findings demonstrate that automated video analysis is an effective and non-invasive tool for collecting detailed traffic data. It offers valuable insights for promoting sustainable mobility, especially cycling, in medium-sized European cities like Maribor. Furthermore, the demonstrated accuracy and scalability of automated classification suggest that this approach can effectively replace manual traffic counts still commonly used in baseline traffic studies, thereby streamlining data collection in transport planning processes. However, the study’s limitations—including its focus on a single morning peak and lack of socio-demographic or trip purpose data—suggest the need for broader spatial and temporal coverage in future research. Integrating this approach with additional data sources could further support evidence-based transport planning and policy development.