Rastreamento em tempo real de múltiplos objetos por associação de detecções

Abstract

With the recent advances in the object detection research field, tracking-by-detection has become the leading paradigm adopted by multi-object tracking algorithms. By extracting different features from detected objects, those algorithms can estimate the objects’ similarities and association patterns along successive frames. However, since similarity functions applied by tracking algorithms are handcrafted, it is difficult to employ them in new contexts. In this study, it is investigated the use of artificial neural networks to automatically learning a similarity function that can be used among detections. During training, the networks were introduced to correct and incorrect association patterns, sampled from real tracking datasets. For such, different motion and appearance features have been considered. A trained network has been inserted into a multiple-object tracking framework, which has been assessed on three different experiment scenarios: the tracking of pedestrians, the tracking of bus passengers and the automatic counting of bus passengers. During the first, conducted on the MOT Challenge benchmark, the proposed method scored an accuracy similar to state-of-the-art trackers, but at a computational cost at least 16 times lower. In the second experiment, which was based on a local dataset, the proposed tracker matched the results obtained by its direct baseline, the DeepSORT tracker, but with a speed gain of 42.8%. Finally, the third experiment was a study case, where the median absolute error scored by the proposed method was 40.7% inferior to its baseline. In the end, this study could demonstrate that the proposal method can be automatically adapted to different tracking scenarios while presenting highly competitive cost-effectiveness when compared to those algorithms considered in the experiments.