Confluence: A Robust Non-IoU Alternative to Non-Maxima Suppression in Object Detection

Andrew J. Shepley, Greg Falzon, Paul Kwan, Ljiljana Brankovic

Research output: Contribution to journalArticlepeer-review

18 Citations (Scopus)

Abstract

Confluence is a novel non-Intersection over Union (IoU) alternative to Non-Maxima Suppression (NMS) in bounding box post-processing in object detection. It overcomes the inherent limitations of IoU-based NMS variants to provide a more stable, consistent predictor of bounding box clustering by using a normalized Manhattan Distance inspired proximity metric to represent bounding box clustering. Unlike Greedy and Soft NMS, it does not rely solely on classification confidence scores to select optimal bounding boxes, instead selecting the box which is closest to every other box within a given cluster and removing highly confluent neighboring boxes. Confluence is experimentally validated on the MS COCO and CrowdHuman benchmarks, improving Average Precision by 0.2 - 2.7% and 1 - 3.8% respectively and Average Recall by 1.3 - 9.3 and 2.4 - 7.3% when compared against Greedy and Soft-NMS variants. Quantitative results are supported by extensive qualitative analysis and threshold sensitivity analysis experiments support the conclusion that Confluence is more robust than NMS variants. Confluence represents a paradigm shift in bounding box processing, with potential to replace IoU in bounding box regression processes.

Original languageEnglish
Pages (from-to)11561-11574
Number of pages14
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume45
Issue number10
DOIs
Publication statusPublished - 1 Oct 2023

Keywords

  • Australia
  • Computer vision
  • Confluence
  • Deep learning
  • Detectors
  • Edge and feature detection
  • Feature representation
  • Image Processing and Computer Vision
  • Location awareness
  • Machine learning
  • Non-Maxima Suppression
  • Object detection
  • Pipelines
  • Standards
  • Task analysis
  • confluence
  • feature representation
  • object detection
  • image processing and computer vision
  • machine learning
  • deep learning
  • edge and feature detection
  • non-maxima suppression

Fingerprint

Dive into the research topics of 'Confluence: A Robust Non-IoU Alternative to Non-Maxima Suppression in Object Detection'. Together they form a unique fingerprint.

Cite this