Fig. 3From: Language agnostic missing subtitle detectionCNNTD-small and CNNTD-large model architecturesBack to article page