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

Precise 6D pose estimation of rigid objects from RGB images is a critical butchallenging task in robotics, augmented reality and human-computer interaction.To address this problem, we propose DeepRM, a novel recurrent networkarchitecture for 6D pose refinement. DeepRM leverages initial coarse poseestimates to render synthetic images of target objects. The rendered images arethen matched with the observed images to predict a rigid transform for updatingthe previous pose estimate. This process is repeated to incrementally refinethe estimate at each iteration. The DeepRM architecture incorporates LSTM unitsto propagate information through each refinement step, significantly improvingoverall performance. In contrast to current 2-stage Perspective-n-Point basedsolutions, DeepRM is trained end-to-end, and uses a scalable backbone that canbe tuned via a single parameter for accuracy and efficiency. During training, amulti-scale optical flow head is added to predict the optical flow between theobserved and synthetic images. Optical flow prediction stabilizes the trainingprocess, and enforces the learning of features that are relevant to the task ofpose estimation. Our results demonstrate that DeepRM achieves state-of-the-artperformance on two widely accepted challenging datasets.

DeepRM: Deep Recurrent Matching for 6D Pose Refinement