Fig. 11From: Human activity prediction using saliency-aware motion enhancement and weighted LSTM networkVisualization of saliency-aware motion enhancement. a Perform well. b Perform poorlyBack to article page