Human trajectory prediction in crowded scene using social-affinity Long Short-Term Memory

Highlights：

• This paper proposes a novel human trajectory prediction model in a crowded scene called the social-affinity LSTM model.

• We formulate the problem of trajectory prediction together with interactions among people as a sequence generation task with social affinity.

• Our model can learn general human mobility patterns and predict individuals trajectories based on their past positions, in particular, with influences of their neighbors in the Social Affinity Map (SAM).

• Our model outperforms the state-of-the-art methods on these datasets with the best results, especially the datasets with more social affinity phenomena.