In open spaces such as parks, public squares or open-air theaters, various types of events can be organized, such as festivals, parties or shows. Depending on the type of event, it may be essential to track the trajectory of each individual present in this space, particularly in a crowd management or security context. Such individual tracking applications could be of great use in ensuring the safety and smooth running of these night-time events. Depending on the type of application, it may also be necessary to track trajectories in real time, using an infrared camera. A number of detection and trajectory tracking algorithms already exist, but not all of them are suitable for the purpose described above. The aim of our work is then to adapt state-of-the-art (SOTA) object detection methods to our situation by making them more robust to crowd detection, and then to combine them with SOTA trackers to carry out a comparative study of the results obtained.