Fig. 7From: Human activity prediction using saliency-aware motion enhancement and weighted LSTM networkAccuracy comparison of different modalities for three datasets. a UT-Interaction set 1. b UT-Interaction set 2. c Sub-JHMDBBack to article page