search

jchengai / forecast-mae

[ICCV'2023] Forecast-MAE: Self-supervised Pre-training for Motion Forecasting with Masked Autoencoders
https://arxiv.org/pdf/2308.09882.pdf
154 stars 16 forks source link

The data #8

Closed kelen1 closed 8 months ago

kelen1 commented 8 months ago

Hey! I'm a beginner in the field of vehicle trajectory prediction, and I would like to ask if I can apply the code to a normal vehicle trajectory, which only contains attributes such as x and y, speed, and lacks elements such as maps. Looking forward to your reply!

jchengai commented 8 months ago

Hi @kelen1, yes, for example, you can remove the lane_feat in the concatenation

https://github.com/jchengai/forecast-mae/blob/cb86ea92601d23a8af7713389e9fb78d7e546a65/src/model/model_forecast.py#L125

Besides, I believe there are several works focusing on map-free prediction (e.g., CART-Pred), and almost all prediction networks can be used for map-free prediction by just droping the map features.

kelen1 commented 8 months ago

I get it. Thank you very much!