Semi-supervised Object Detection via VC Learning: Computer Vision, ECCV 2022, Pt. XXXI

CR Chen, Kurt Debattista, Jiwan Han, S Avidan (Editor), G Brostow (Editor), M Cisse (Editor), GM Farinella (Editor), T Hassner (Editor)

Research output: Chapter in Book/Report/Conference proceedingConference Proceeding (Non-Journal item)

2 Citations (SciVal)

Abstract

Due to the costliness of labelled data in real-world applications, semi-supervised object detectors, underpinned by pseudo labelling, are appealing. However, handling confusing samples is nontrivial: discarding valuable confusing samples would compromise the model generalisation while using them for training would exacerbate the confirmation bias issue caused by inevitable mislabelling. To solve this problem, this paper proposes to use confusing samples proactively without label correction. Specifically, a virtual category (VC) is assigned to each confusing sample such that they can safely contribute to the model optimisation even without a concrete label. It is attributed to specifying the embedding distance between the training sample and the virtual category as the lower bound of the inter-class distance. Moreover, we also modify the localisation loss to allow high-quality boundaries for location regression. Extensive experiments demonstrate that the proposed VC learning significantly surpasses the state-of-the-art, especially with small amounts of available labels.
Original languageEnglish
Title of host publicationComputer Vision – ECCV 2022
EditorsShai Avidan, Gabriel Brostow, Moustapha Cissé, Giovanni Maria Farinella, Tal Hassner
Pages169-185
Number of pages17
DOIs
Publication statusPublished - 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13691 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Keywords

  • Semi-supervised learning
  • Object detection

Fingerprint

Dive into the research topics of 'Semi-supervised Object Detection via VC Learning: Computer Vision, ECCV 2022, Pt. XXXI'. Together they form a unique fingerprint.

Cite this