Fig. 2From: Robust visual tracking using very deep generative modelThe architecture of the feature extraction stageBack to article page