search

ZikangZhou / HiVT

[CVPR 2022] HiVT: Hierarchical Vector Transformer for Multi-Agent Motion Prediction
https://openaccess.thecvf.com/content/CVPR2022/papers/Zhou_HiVT_Hierarchical_Vector_Transformer_for_Multi-Agent_Motion_Prediction_CVPR_2022_paper.pdf
Apache License 2.0
577 stars 115 forks source link

Question Regarding to Temporal Encoder #23

Closed SwagJ closed 1 year ago

SwagJ commented 1 year ago

Hi @ZikangZhou,

There is one question regarding to the temporal encoder layer. When computing the attention in temporal encoder layer, all the agent within the batch is calculated instead of batch-wise attention, I was wondering fusing such batched information? Thank you in advance.

Best,

ZikangZhou commented 1 year ago

The temporal encoder performs attention across different time steps within the same agent, so there isn't any information fusion across different agents.