Urban environments are still a challenge for Autonomous Vehicles, due to strong interactions with other vehicles and pedestrians. Machine learning methods are increasingly explored to tackle these situations, but their performances are highly conditioned on the availability of vehicle trajectory datasets. Such datasets are usually created with specific infrastructures and equipments, which is costly and not scalable. As a result, only a few datasets of vehicle trajectories are currently available, representing very specific situations such as highway driving, and containing a limited number of trajectories. We argue that to unleash the potential of behavior learning methods for autonomous vehicles, we need large datasets of accurate vehicle trajectories, representing interacting vehicles in very diverse situations. In this paper, we introduce a fully automatic and scalable framework for accurate vehicle trajectory extraction from single fixed monocular traffic cameras. We leverage the fact that traffic cameras represent a very large and cost-effective source of highly diverse vehicle trajectories, as they are generally located at places where traffic is dense and where a lot of interactions occur (e.g intersections). With this work, we aim at developing a framework for effortless and accurate vehicle trajectory dataset creation at large-scale. Furthermore, we provide an open-source implementation of this framework https://github.com/AubreyC/trajectory-extractor.