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

Recent advancements in energy-efficient hardware technology is driving theexponential growth we are experiencing in the Internet of Things (IoT) space,with more pervasive computations being performed near to data generationsources. A range of intelligent devices and applications performing localdetection is emerging (activity recognition, fitness monitoring, etc.) bringingwith them obvious advantages such as reducing detection latency for improvedinteraction with devices and safeguarding user data by not leaving the device.Video processing holds utility for many emerging applications and datalabelling in the IoT space. However, performing this video processing with deepneural networks at the edge of the Internet is not trivial. In this paper weshow that pedestrian location estimation using deep neural networks isachievable on fixed cameras with limited compute resources. Our approach usespose estimation from key body points detection to extend pedestrian skeletonwhen whole body not in image (occluded by obstacles or partially outside offrame), which achieves better location estimation performance (infrence timeand memory footprint) compared to fitting a bounding box over pedestrian andscaling. We collect a sizable dataset comprising of over 2100 frames in videosfrom one and two surveillance cameras pointing from different angles at thescene, and annotate each frame with the exact position of person in image, in42 different scenarios of activity and occlusion. We compare our poseestimation based location detection with a popular detection algorithm, YOLOv2,for overlapping bounding-box generation, our solution achieving fasterinference time (15x speedup) at half the memory footprint, within resourcecapabilities on embedded devices, which demonstrate that CamLoc is an efficientsolution for location estimation in videos on smart-cameras.

CamLoc: Pedestrian Location Detection from Pose Estimation on Resource-constrained Smart-cameras