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

Although human pose estimation for various computer vision (CV) applicationshas been studied extensively in the last few decades, yet in-bed poseestimation using camera-based vision methods has been ignored by the CVcommunity because it is assumed to be identical to the general purpose poseestimation methods. However, in-bed pose estimation has its own specializedaspects and comes with specific challenges including the notable differences inlighting conditions throughout a day and also having different posedistribution from the common human surveillance viewpoint. In this paper, wedemonstrate that these challenges significantly lessen the effectiveness ofexisting general purpose pose estimation models. In order to address thelighting variation challenge, infrared selective (IRS) image acquisitiontechnique is proposed to provide uniform quality data under various lightingconditions. In addition, to deal with unconventional pose perspective, a 2-endhistogram of oriented gradient (HOG) rectification method is presented. In thiswork, we explored the idea of employing a pre-trained convolutional neuralnetwork (CNN) model trained on large public datasets of general human poses andfine-tuning the model using our own shallow in-bed IRS dataset. We developed anIRS imaging system and collected IRS image data from several realisticlife-size mannequins in a simulated hospital room environment. A pre-trainedCNN called convolutional pose machine (CPM) was repurposed for in-bed poseestimation by fine-tuning its specific intermediate layers. Using the HOGrectification method, the pose estimation performance of CPM significantlyimproved by 26.4% in PCK0.1 criteria compared to the model without suchrectification.

In-Bed Pose Estimation: Deep Learning with Shallow Dataset