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 scene reconstruction from posed images is challenging due to thecomplexity of a large environment. Recent volumetric methods learn to directlypredict the TSDF volume and have demonstrated promising results in this task.However, most methods focus on how to extract and fuse the 2D features to a 3Dfeature volume, but none of them improve the way how the 3D volume isaggregated. In this work, we propose an SDF transformer network, which replacesthe role of 3D CNN for better 3D feature aggregation. To reduce the explosivecomputation complexity of the 3D multi-head attention, we propose a sparsewindow attention module, where the attention is only calculated between thenon-empty voxels within a local window. Then a top-down-bottom-up 3D attentionnetwork is built for 3D feature aggregation, where a dilate-attention structureis proposed to prevent geometry degeneration, and two global modules areemployed to equip with global receptive fields. The experiments on multipledatasets show that this 3D transformer network generates a more accurate andcomplete reconstruction, which outperforms previous methods by a large margin.Remarkably, the mesh accuracy is improved by 41.8%, and the mesh completenessis improved by 25.3% on the ScanNet dataset. Project page:https://weihaosky.github.io/sdfformer.

3D Former: Monocular Scene Reconstruction with 3D SDF Transformers