English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Sparse algorithms offer great flexibility for multi-view temporal perceptiontasks. In this paper, we present an enhanced version of Sparse4D, in which weimprove the temporal fusion module by implementing a recursive form ofmulti-frame feature sampling. By effectively decoupling image features andstructured anchor features, Sparse4D enables a highly efficient transformationof temporal features, thereby facilitating temporal fusion solely through theframe-by-frame transmission of sparse features. The recurrent temporal fusionapproach provides two main benefits. Firstly, it reduces the computationalcomplexity of temporal fusion from $O(T)$ to $O(1)$, resulting in significantimprovements in inference speed and memory usage. Secondly, it enables thefusion of long-term information, leading to more pronounced performanceimprovements due to temporal fusion. Our proposed approach, Sparse4Dv2, furtherenhances the performance of the sparse perception algorithm and achievesstate-of-the-art results on the nuScenes 3D detection benchmark. Code will beavailable at \url{https://github.com/linxuewu/Sparse4D}.

Sparse4D v2: Recurrent Temporal Fusion with Sparse Model