search

tanjatang / DCENet

Exploring Dynamic Context for Multi-path Trajectory Prediction
24 stars 17 forks source link

Results of using pretrained model named best.hdf5 #4

Closed laurelyi2020 closed 2 years ago

laurelyi2020 commented 3 years ago

Hello, when applying the pretrained model best.hdf5, the test performance of bookstore-3 is shown as below: image it doesn't look like the one in the paper, all the prediction lines are just straight, I am wondering about it, could you please give me some advice on how it's happening? Thx and wish for you reply.

haohao11 commented 2 years ago

Hi, thank you very much for the question. One can also play around with the hyper-parameters --num_pred and --beta to get more divergent multi-path predictions.