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 depth estimation plays a critical role in various computer visionand robotics applications such as localization, mapping, and 3D objectdetection. Recently, learning-based algorithms achieve huge success in depthestimation by training models with a large amount of data in a supervisedmanner. However, it is challenging to acquire dense ground truth depth labelsfor supervised training, and the unsupervised depth estimation using monocularsequences emerges as a promising alternative. Unfortunately, most studies onunsupervised depth estimation explore loss functions or occlusion masks, andthere is little change in model architecture in that ConvNet-basedencoder-decoder structure becomes a de-facto standard for depth estimation. Inthis paper, we employ a convolution-free Swin Transformer as an image featureextractor so that the network can capture both local geometric features andglobal semantic features for depth estimation. Also, we propose a DenselyCascaded Multi-scale Network (DCMNet) that connects every feature map directlywith another from different scales via a top-down cascade pathway. This denselycascaded connectivity reinforces the interconnection between decoding layersand produces high-quality multi-scale depth outputs. The experiments on twodifferent datasets, KITTI and Make3D, demonstrate that our proposed methodoutperforms existing state-of-the-art unsupervised algorithms.

SwinDepth: Unsupervised Depth Estimation using Monocular Sequences via Swin Transformer and Densely Cascaded Network