Rethinking Self-training for Semi-supervised Landmark Detection: A Selection-free Approach

arXiv:2404.04556v1 Announce Type: new
Abstract: Self-training is a simple yet effective method for semi-supervised learning, during which pseudo-label selection plays an important role for handling confirmation bias. Despite its popularity, applying self-training to landmark detection faces three problems: 1) The selected confident pseudo-labels often contain data bias, which may hurt model performance; 2) It is not easy to decide a proper threshold for sample selection as the localization task can be sensitive to noisy pseudo-labels; 3) coordinate regression does not …

abstract arxiv bias confirmation bias cs.cv data data bias detection free labels landmark role self-training semi-supervised semi-supervised learning simple supervised learning training type