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

LiDAR and cameras are complementary sensors for 3D object detection inautonomous driving. However, it is challenging to explore the unnaturalinteraction between point clouds and images, and the critical factor is how toconduct feature alignment of heterogeneous modalities. Currently, many methodsachieve feature alignment by projection calibration only, without consideringthe problem of coordinate conversion accuracy errors between sensors, leadingto sub-optimal performance. In this paper, we present GraphAlign, a moreaccurate feature alignment strategy for 3D object detection by graph matching.Specifically, we fuse image features from a semantic segmentation encoder inthe image branch and point cloud features from a 3D Sparse CNN in the LiDARbranch. To save computation, we construct the nearest neighbor relationship bycalculating Euclidean distance within the subspaces that are divided into thepoint cloud features. Through the projection calibration between the image andpoint cloud, we project the nearest neighbors of point cloud features onto theimage features. Then by matching the nearest neighbors with a single pointcloud to multiple images, we search for a more appropriate feature alignment.In addition, we provide a self-attention module to enhance the weights ofsignificant relations to fine-tune the feature alignment between heterogeneousmodalities. Extensive experiments on nuScenes benchmark demonstrate theeffectiveness and efficiency of our GraphAlign.

GraphAlign: Enhancing Accurate Feature Alignment by Graph matching for Multi-Modal 3D Object Detection