XiaolongTang23 / HPNet

[CVPR 2024] HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention
Apache License 2.0
96 stars 11 forks source link

Is there any mistakes in t2m_heading_t? #10

Closed ilzkl1991 closed 1 month ago

ilzkl1991 commented 1 month ago

Is there any mistakes in t2m_heading_t? maybe it should be data['agent']['heading'][:,:self.num_historical_steps].reshape(-1)?

t2m edge

    t2m_position_t = data['agent']['position'][:,:self.num_historical_steps].reshape(-1,2)      #[(N1,...,Nb)*H,2]
    t2m_position_m = m_position.reshape(-1,2)                                                   #[(N1,...,Nb)*H*K,2]
    t2m_heading_t = data['agent']['heading'].reshape(-1)                                        #[(N1,...,Nb)]
    t2m_heading_m = m_heading.reshape(-1)                                                       #[(N1,...,Nb)*H*K]
    t2m_valid_mask_t = data['agent']['visible_mask'][:,:self.num_historical_steps]              #[(N1,...,Nb),H]
    t2m_valid_mask_m = m_valid_mask.reshape(num_all_agent,-1)                                   #[(N1,...,Nb),H*K]
    t2m_valid_mask = t2m_valid_mask_t.unsqueeze(2) & t2m_valid_mask_m.unsqueeze(1)              #[(N1,...,Nb),H,H*K]