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

Neural Radiance Fields (NeRF) have demonstrated impressive potential insynthesizing novel views from dense input, however, their effectiveness ischallenged when dealing with sparse input. Existing approaches that incorporateadditional depth or semantic supervision can alleviate this issue to an extent.However, the process of supervision collection is not only costly but alsopotentially inaccurate, leading to poor performance and generalization abilityin diverse scenarios. In our work, we introduce a novel model: theCollaborative Neural Radiance Fields (ColNeRF) designed to work with sparseinput. The collaboration in ColNeRF includes both the cooperation betweensparse input images and the cooperation between the output of the neuralradiation field. Through this, we construct a novel collaborative module thataligns information from various views and meanwhile imposes self-supervisedconstraints to ensure multi-view consistency in both geometry and appearance. ACollaborative Cross-View Volume Integration module (CCVI) is proposed tocapture complex occlusions and implicitly infer the spatial location ofobjects. Moreover, we introduce self-supervision of target rays projected inmultiple directions to ensure geometric and color consistency in adjacentregions. Benefiting from the collaboration at the input and output ends,ColNeRF is capable of capturing richer and more generalized scenerepresentation, thereby facilitating higher-quality results of the novel viewsynthesis. Extensive experiments demonstrate that ColNeRF outperformsstate-of-the-art sparse input generalizable NeRF methods. Furthermore, ourapproach exhibits superiority in fine-tuning towards adapting to new scenes,achieving competitive performance compared to per-scene optimized NeRF-basedmethods while significantly reducing computational costs. Our code is availableat: https://github.com/eezkni/ColNeRF.

ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field