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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Recovery of articulated 3D structure from 2D observations is a challengingcomputer vision problem with many applications. Current learning-basedapproaches achieve state-of-the-art accuracy on public benchmarks but arerestricted to specific types of objects and motions covered by the trainingdatasets. Model-based approaches do not rely on training data but show loweraccuracy on these datasets. In this paper, we introduce a model-based methodcalled Structure from Articulated Motion (SfAM), which can recover multipleobject and motion types without training on extensive data collections. At thesame time, it performs on par with learning-based state-of-the-art approacheson public benchmarks and outperforms previous non-rigid structure from motion(NRSfM) methods. SfAM is built upon a general-purpose NRSfM technique whileintegrating a soft spatio-temporal constraint on the bone lengths. We usealternating optimization strategy to recover optimal geometry (i.e., boneproportions) together with 3D joint positions by enforcing the bone lengthsconsistency over a series of frames. SfAM is highly robust to noisy 2Dannotations, generalizes to arbitrary objects and does not rely on trainingdata, which is shown in extensive experiments on public benchmarks and realvideo sequences. We believe that it brings a new perspective on the domain ofmonocular 3D recovery of articulated structures, including human motioncapture.

Structure from Articulated Motion: Accurate and Stable Monocular 3D Reconstruction without Training Data