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/

Zendy Plus

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

Zendy Tools

Zendy Open

Commercial depth sensors usually generate noisy and missing depths,especially on specular and transparent objects, which poses critical issues todownstream depth or point cloud-based tasks. To mitigate this problem, wepropose a powerful RGBD fusion network, SwinDRNet, for depth restoration. Wefurther propose Domain Randomization-Enhanced Depth Simulation (DREDS) approachto simulate an active stereo depth system using physically based rendering andgenerate a large-scale synthetic dataset that contains 130K photorealistic RGBimages along with their simulated depths carrying realistic sensor noises. Toevaluate depth restoration methods, we also curate a real-world dataset, namelySTD, that captures 30 cluttered scenes composed of 50 objects with differentmaterials from specular, transparent, to diffuse. Experiments demonstrate thatthe proposed DREDS dataset bridges the sim-to-real domain gap such that,trained on DREDS, our SwinDRNet can seamlessly generalize to other real depthdatasets, e.g. ClearGrasp, and outperform the competing methods on depthrestoration with a real-time speed. We further show that our depth restorationeffectively boosts the performance of downstream tasks, includingcategory-level pose estimation and grasping tasks. Our data and code areavailable at https://github.com/PKU-EPIC/DREDS

Domain Randomization-Enhanced Depth Simulation and Restoration for Perceiving and Grasping Specular and Transparent Objects