This study explores the influence of a high percentage of motorcycles on the traffic flow and congestion in Marrakech by examining the impact of motorcycle positioning in shaping urban traffic dynamics, in particular, the start-up lost time at signalized intersections. Different motorcycle positioning strategies are analyzed to improve intersection efficiency and safety. A twofold approach was followed to achieve this objective. First, empirical data were collected using computer vision techniques. Second, different strategies were simulated in VISSIM based on the collected data. The approach adopted for data collection was based on mobile phone video recording at a representative signalized intersection in Marrakech, capturing traffic behaviors during four distinct time periods. Then the YOLOv8 algorithm was employed for real-time object detection and analysis, allowing precise monitoring of motorcycle positioning and examining its influence on the start-up lost time. Afterwards, VISSIM simulations were implemented, on the basis of the collected data, to explore various scenarios, such as motorcycles sharing lanes with cars or dedicated motorcycle lanes. The results reveal a compelling correlation between motorcycle proximity to cars and traffic congestion, with closer positioning leading to increased congestion, longer travel times, reduced average vehicle speeds, and extended queue lengths at intersections. On the contrary, scenarios with dedicated motorcycle lanes consistently show reduced congestion and smoother traffic flow.