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

Monocular 3D object detection aims to locate objects in different scenes withjust a single image. Due to the absence of depth information, several monocular3D detection techniques have emerged that rely on auxiliary depth maps from thedepth estimation task. There are multiple approaches to understanding therepresentation of depth maps, including treating them as pseudo-LiDAR pointclouds, leveraging implicit end-to-end learning of depth information, orconsidering them as an image input. However, these methods have certaindrawbacks, such as their reliance on the accuracy of estimated depth maps andsuboptimal utilization of depth maps due to their image-based nature. WhileLiDAR-based methods and convolutional neural networks (CNNs) can be utilizedfor pseudo point clouds and depth maps, respectively, it is always analternative. In this paper, we propose a framework named the Adaptive DistanceInterval Separation Network (ADISN) that adopts a novel perspective onunderstanding depth maps, as a form that lies between LiDAR and images. Weutilize an adaptive separation approach that partitions the depth map intovarious subgraphs based on distance and treats each of these subgraphs as anindividual image for feature extraction. After adaptive separations, eachsubgraph solely contains pixels within a learned interval range. If there is atruncated object within this range, an evident curved edge will appear, whichwe can leverage for texture extraction using CNNs to obtain rich depthinformation in pixels. Meanwhile, to mitigate the inaccuracy of depthestimation, we designed an uncertainty module. To take advantage of both imagesand depth maps, we use different branches to learn localization detection tasksand appearance tasks separately.

Understanding Depth Map Progressively: Adaptive Distance Interval Separation for Monocular 3d Object Detection