Transformer Networks for Trajectory Forecasting

18 Mar 2020Francesco GiuliariIrtiza HasanMarco CristaniFabio Galasso

Most recent successes on forecasting the people motion are based on LSTM models and all most recent progress has been achieved by modelling the social interaction among people and the people interaction with the scene. We question the use of the LSTM models and propose the novel use of Transformer Networks for trajectory forecasting... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Trajectory Prediction ETH/UCY Tranformer TF ADE-8/12 0.31 # 1

Methods used in the Paper