Open coco-99-coco opened 3 years ago
Hi @coco-99-coco ,
I'm not sure I fully understand your question. The trajectories in Argoverse 1 are organized in scenarios with dozens of objects sampled at the same time and the same place (within 100 meters of the autonomous vehicle, roughly). Argoverse 1 doesn't have category labels for those objects, though. You can't tell if an object is a pedestrian or a vehicle. Although most things moving quickly on the road will be vehicles, of course.
The Argoverse 2 motion forecasting dataset will contain the category of every actor in the scene. We hope to release it soon. It is described here https://openreview.net/forum?id=vKQGe36av4k
A common scenario is that there are multiple interactive objects on a road at the same time, such as vehicles and pedestrians. Is there a label for the samples sampled at the same time in the same place in the training data set?