[ECCV 2022] VL-LTR: Learning Class-wise Visual-Linguistic Representation for LTR
ContentsIntroductionMethodologyClass-wiseVisual-LinguisticPre-training(CVLP)Language-GuidedRecognitionExperimentsReferencesIntroduction作者将大规模预训练多模态模型(i.e.,CLIP)用在了长尾分类任务上