search

cunjunyu / STAR

[ECCV 2020] Code for "Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Prediction"
MIT License
338 stars 78 forks source link

DataSet #11

Closed MazenHamdi closed 3 years ago

MazenHamdi commented 3 years ago

Hello, I loved the work, very well done,

I am not familiar with the dataSets that you used, so I tried to visualize it, I couldn't identify the pedestrian from the trajectory, I noticed in the head of the dataset number keep repeating, are they pedestrian indexes ? and the other rows, are they x and y of the trajectory ? Thanks a lot

cunjunyu commented 3 years ago
Screen Shot 2021-06-06 at 2 03 53 PM

First Row: Timestamp Second Row: Pedestrian Indices Third Row: X coordinate Fourth Row: Y coordinate

Thanks.

cunjunyu commented 3 years ago

I will close the issue for now. Feel free to reopen it if you have any further questions.

MazenHamdi commented 3 years ago

What did you use the functions (find_trajectory_fragment, get_seq_from_index_balance, load_dict .etc) ? they are not used in the training nor the test, I deleted them and the model works just fine.

cunjunyu commented 3 years ago

They are used to generate training/testing datasets. The model still works because you have already generated the pickle file of the dataset when you first ran the code. see here https://github.com/Majiker/STAR/blob/d2dd113e3deb1098760c454ab15e7cb0652f3c0c/src/utils.py#L65

Thank you.

MazenHamdi commented 3 years ago

About the dataset, is the pedestrian in hotel with index 1 the same pedestrian in the univ with the same index ?

cunjunyu commented 3 years ago

No, they are not. Hotel and Univ are two datasets collected at different places on different dates.

MazenHamdi commented 3 years ago

what exactly do you mean by "window" ? (sorry for all these questions, I am really interested in this work and I want to understand every single line of code)

cunjunyu commented 3 years ago

'window' refers to a time period which is normally 20 frames (8 for observation and 12 for prediction)