In recent years, there is a shift from modeling the tracking problem based on Bayesian formulation towards using deep neural networks. Towards this end, in this paper the effectiveness of various deep neural networks for predicting future pedestrian paths are evaluated. The analyzed deep networks solely rely, like in the traditional approaches, on observed tracklets without human-human interaction information. The evaluation is done on the publicly available TrajNet benchmark dataset [39], which builds up a repository of considerable and popular datasets for trajectory prediction. We show how a Recurrent-Encoder with a Dense layer stacked on top, referred to as RED-predictor, is able to achieve toprank at the TrajNet 2018 challenge compared to elaborated models. Further, we investigate failure cases and give explanations for observed phenomena, and give some recommendations for overcoming demonstrated shortcomings.

Recurrent neural networks are able to learn complex long-term relationships from sequential data and output a probability density function over the state space. Therefore, recurrent models are a natural choice to address path prediction tasks, where a trained model is used to generate future expectations from past observations. When applied to security applications, like predicting pedestrian paths for risk assessment, a point-wise greedy evaluation of the output pdf is not feasible, since the environment often allows multiple choices. Therefore, a robust risk assessment has to take all options into account, even if they are overall not very likely.Towards this end, a combination of particle filtering strategies and a LSTM-MDL model is proposed to address a multimodal path prediction task. The capabilities and viability of the proposed approach are evaluated on several synthetic test conditions, yielding the counter-intuitive result that the simplest approach performs best. Further, the feasibility of the proposed approach is illustrated on several real world scenes.

Recurrent neural networks, like the LSTM model, have been applied to various sequence learning tasks with great success. Following this, it seems natural to use LSTM models for predicting future locations in object tracking tasks. In this paper, we evaluate an adaption of a LSTM-MDL model and investigate its reliability in the context of pedestrian trajectory prediction. Thereby, we demonstrate the fallacy of solely relying on prediction metrics for evaluating the model and how the models capabilities can lead to suboptimal prediction results. Towards this end, two experiments are provided. Firstly, the models prediction abilities are evaluated on publicly available surveillance datasets. Secondly, the capabilities of capturing motion patterns are examined. Further, we investigate failure cases and give explanations for observed phenomena, granting insight into the models reliability in tracking applications. Lastly, we give some hints how demonstrated shortcomings may be circumvented.

No abstract

No abstract