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

Unsupervised learning of depth and ego-motion from unlabelled monocularvideos has recently drawn great attention, which avoids the use of expensiveground truth in the supervised one. It achieves this by using the photometricerrors between the target view and the synthesized views from its adjacentsource views as the loss. Despite significant progress, the learning stillsuffers from occlusion and scene dynamics. This paper shows that carefullymanipulating photometric errors can tackle these difficulties better. Theprimary improvement is achieved by a statistical technique that can mask outthe invisible or nonstationary pixels in the photometric error map and thusprevents misleading the networks. With this outlier masking approach, the depthof objects moving in the opposite direction to the camera can be estimated moreaccurately. To the best of our knowledge, such scenarios have not beenseriously considered in the previous works, even though they pose a higher riskin applications like autonomous driving. We also propose an efficient weightedmulti-scale scheme to reduce the artifacts in the predicted depth maps.Extensive experiments on the KITTI dataset show the effectiveness of theproposed approaches. The overall system achieves state-of-theart performance onboth depth and ego-motion estimation.

DiPE: Deeper into Photometric Errors for Unsupervised Learning of Depth and Ego-motion from Monocular Videos