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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Recent work has shown that convolutional neural networks (CNNs) can beapplied successfully in disparity estimation, but these methods still sufferfrom errors in regions of low-texture, occlusions and reflections.Concurrently, deep learning for semantic segmentation has shown great progressin recent years. In this paper, we design a CNN architecture that combinesthese two tasks to improve the quality and accuracy of disparity estimationwith the help of semantic segmentation. Specifically, we propose a networkstructure in which these two tasks are highly coupled. One key novelty of thisapproach is the two-stage refinement process. Initial disparity estimates arerefined with an embedding learned from the semantic segmentation branch of thenetwork. The proposed model is trained using an unsupervised approach, in whichimages from one half of the stereo pair are warped and compared against imagesfrom the other camera. Another key advantage of the proposed approach is that asingle network is capable of outputting disparity estimates and semanticlabels. These outputs are of great use in autonomous vehicle operation; withreal-time constraints being key, such performance improvements increase theviability of driving applications. Experiments on KITTI and Cityscapes datasetsshow that our model can achieve state-of-the-art results and that leveragingembedding learned from semantic segmentation improves the performance ofdisparity estimation.

DispSegNet: Leveraging Semantics for End-to-End Learning of Disparity Estimation from Stereo Imagery