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

Object detectors are at the heart of many semi- and fully autonomous decisionsystems and are poised to become even more indispensable. They are, however,still lacking in accessibility and can sometimes produce unreliablepredictions. Especially concerning in this regard are the -- essentiallyhand-crafted -- non-maximum suppression algorithms that lead to an obfuscatedprediction process and biased confidence estimates. We show that we caneliminate classic NMS-style post-processing by using IoU-aware calibration.IoU-aware calibration is a conditional Beta calibration; this makes itparallelizable with no hyper-parameters. Instead of arbitrary cutoffs ordiscounts, it implicitly accounts for the likelihood of each detection being aduplicate and adjusts the confidence score accordingly, resulting inempirically based precision estimates for each detection. Our extensiveexperiments on diverse detection architectures show that the proposed IoU-awarecalibration can successfully model duplicate detections and improvecalibration. Compared to the standard sequential NMS and calibration approach,our joint modeling can deliver performance gains over the best NMS-basedalternative while producing consistently better-calibrated confidencepredictions with less complexity. The\hyperlink{https://github.com/Blueblue4/IoU-AwareCalibration}{code} for all ourexperiments is publicly available.

Do We Still Need Non-Maximum Suppression? Accurate Confidence Estimates and Implicit Duplication Modeling with IoU-Aware Calibration