Résumé
This paper introduces FORT-RAJ, a hybrid model designed forpedestrian trajectory prediction in the context of top-view fisheye images. Toachieve this, FORT-RAJ merges the FORT (Fisheye Online Realtime Tracking)algorithm, which tracks people using fisheye cameras without predictioncapabilities, with the GATraj model, known for trajectory prediction but notyet adapted for fisheye images. The proposed method, FORT-RAJ, is designedto detect pedestrians, track their trajectories, and predict their futurepositions. It leverages the wide field of view of fisheye cameras whileaddressing the distortions inherent in such images. The experimentsdemonstrated that the FORT-RAJ model performs satisfactorily on fisheyeimages, achieving an Average Displacement Error (ADE) of 0.38 meters andan Final Displacement Error (FDE) of 0.42 meters.