Search Results for "cross"

...In a nutshell, BEVFormer exploits both spatial and temporal information by interacting with spatial and temporal space through predefined grid-shaped BEV queries. To aggregate spatial information, we design spatial cross-attention that each BEV query extracts the spatial features from the regions of interest across camera views. For temporal information, we propose temporal self-attention to recurrently fuse the history BEV information. Our approach achieves the new state-of-the-art 56.9\% in terms of NDS metric on the nuScenes \texttt{test} set, which is 9.0 points higher than previous best arts and on par with the performance of LiDAR-based baseline.

A new algorithm for extracting roof tops was developed. Using the assumption that roof tops are planar in construction, a new approach was developed using volume of point clouds to determine whether a cluster contains planar points. This approach yields very promising results and with attention applied to its weaknesses, should provide another algorithm which can rival currently available roof top detection methods.