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

The exploration of mutual-benefit cross-domains has shown great potentialtoward accurate self-supervised depth estimation. In this work, we revisitfeature fusion between depth and semantic information and propose an efficientlocal adaptive attention method for geometric aware representation enhancement.Instead of building global connections or deforming attention across thefeature space without restraint, we bound the spatial interaction within alearnable region of interest. In particular, we leverage geometric cues fromsemantic information to learn local adaptive bounding boxes to guideunsupervised feature aggregation. The local areas preclude most irrelevantreference points from attention space, yielding more selective feature learningand faster convergence. We naturally extend the paradigm into a multi-head andhierarchic way to enable the information distillation in different semanticlevels and improve the feature discriminative ability for fine-grained depthestimation. Extensive experiments on the KITTI dataset show that our proposedmethod establishes a new state-of-the-art in self-supervised monocular depthestimation task, demonstrating the effectiveness of our approach over formerTransformer variants.

ROIFormer: Semantic-Aware Region of Interest Transformer for Efficient Self-Supervised Monocular Depth Estimation