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

Surface reconstruction from point clouds is vital for 3D computer vision.State-of-the-art methods leverage large datasets to first learn local contextpriors that are represented as neural network-based signed distance functions(SDFs) with some parameters encoding the local contexts. To reconstruct asurface at a specific query location at inference time, these methods thenmatch the local reconstruction target by searching for the best match in thelocal prior space (by optimizing the parameters encoding the local context) atthe given query location. However, this requires the local context prior togeneralize to a wide variety of unseen target regions, which is hard toachieve. To resolve this issue, we introduce Predictive Context Priors bylearning Predictive Queries for each specific point cloud at inference time.Specifically, we first train a local context prior using a large point clouddataset similar to previous techniques. For surface reconstruction at inferencetime, however, we specialize the local context prior into our PredictiveContext Prior by learning Predictive Queries, which predict adjusted spatialquery locations as displacements of the original locations. This leads to aglobal SDF that fits the specific point cloud the best. Intuitively, the queryprediction enables us to flexibly search the learned local context prior overthe entire prior space, rather than being restricted to the fixed querylocations, and this improves the generalizability. Our method does not requireground truth signed distances, normals, or any additional procedure of signeddistance fusion across overlapping regions. Our experimental results in surfacereconstruction for single shapes or complex scenes show significantimprovements over the state-of-the-art under widely used benchmarks.

Surface Reconstruction from Point Clouds by Learning Predictive Context Priors