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

Self-supervised depth estimation has drawn much attention in recent years asit does not require labeled data but image sequences. Moreover, it can beconveniently used in various applications, such as autonomous driving,robotics, realistic navigation, and smart cities. However, extracting globalcontextual information from images and predicting a geometrically natural depthmap remain challenging. In this paper, we present DLNet for pixel-wise depthestimation, which simultaneously extracts global and local features with theaid of our depth Linformer block. This block consists of the Linformer andinnovative soft split multi-layer perceptron blocks. Moreover, athree-dimensional geometry smoothness loss is proposed to predict ageometrically natural depth map by imposing the second-order smoothnessconstraint on the predicted three-dimensional point clouds, thereby realizingimproved performance as a byproduct. Finally, we explore the multi-scaleprediction strategy and propose the maximum margin dual-scale predictionstrategy for further performance improvement. In experiments on the KITTI andMake3D benchmarks, the proposed DLNet achieves performance competitive to thoseof the state-of-the-art methods, reducing time and space complexities by morethan $62\%$ and $56\%$, respectively. Extensive testing on various real-worldsituations further demonstrates the strong practicality and generalizationcapability of the proposed model.

Self-supervised Depth Estimation Leveraging Global Perception and Geometric Smoothness Using On-board Videos