UniFusion: Unified Multi-view Fusion Transformer for Spatial-Temporal Representation in Bird’s-Eye-View

City-based split

This repo contains the "city-based" split proposed in the paper.

• Trianing set: the samples collected in singapore-queenstown and singapore-hollandvillage.
• Validation set: the samples collected in singapore-onenorth and boston-seaport.

The sizes of training and validation sets of "city-based" split are 26,093 and 8,056. For comparision, the sizes of the vanilla training and validation sets are 28,130 and 6,019.

We also provide a nuscenes_city_based_split.py to generate the city-based split.

Demo

We also provide a demo of our method.