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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

We propose a novel deep reinforcement learning-based approach for 3D objectreconstruction from monocular images. Prior works that use mesh representationsare template based. Thus, they are limited to the reconstruction of objectsthat have the same topology as the template. Methods that use volumetric gridsas intermediate representations are computationally expensive, which limitstheir application in real-time scenarios. In this paper, we propose a novelend-to-end method that reconstructs 3D objects of arbitrary topology from amonocular image. It is composed of of (1) a Vertex Generation Network (VGN),which predicts the initial 3D locations of the object's vertices from an inputRGB image, (2) a differentiable triangulation layer, which learns in anon-supervised manner, using a novel reinforcement learning algorithm, the besttriangulation of the object's vertices, and finally, (3) a hierarchical meshrefinement network that uses graph convolutions to refine the initial mesh. Ourkey contribution is the learnable triangulation process, which recovers in anunsupervised manner the topology of the input shape. Our experiments onShapeNet and Pix3D benchmarks show that the proposed method outperforms thestate-of-the-art in terms of visual quality, reconstruction accuracy, andcomputational time.

Learnable Triangulation for Deep Learning-based 3D Reconstruction of Objects of Arbitrary Topology from Single RGB Images