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

Accurate pedestrian classification and localization have receivedconsiderable attention due to their wide applications such as securitymonitoring, autonomous driving, etc. Although pedestrian detectors have madegreat progress in recent years, the fixed Intersection over Union (IoU) basedassignment-regression manner still limits their performance. Two main factorsare responsible for this: 1) the IoU threshold faces a dilemma that a lower onewill result in more false positives, while a higher one will filter out thematched positives; 2) the IoU-based GT-Proposal assignment suffers from theinconsistent supervision problem that spatially adjacent proposals with similarfeatures are assigned to different ground-truth boxes, which means some verysimilar proposals may be forced to regress towards different targets, and thusconfuses the bounding-box regression when predicting the location results. Inthis paper, we first put forward the question that \textbf{RegressionDirection} would affect the performance for pedestrian detection. Consequently,we address the weakness of IoU by introducing one geometric sensitive searchalgorithm as a new assignment and regression metric. Different from theprevious IoU-based \textbf{one-to-one} assignment manner of one proposal to oneground-truth box, the proposed method attempts to seek a reasonable matchingbetween the sets of proposals and ground-truth boxes. Specifically, we boostthe MR-FPPI under R$_{75}$ by 8.8\% on Citypersons dataset. Furthermore, byincorporating this method as a metric into the state-of-the-art pedestriandetectors, we show a consistent improvement.

Which to Match? Selecting Consistent GT-Proposal Assignment for Pedestrian Detection