Joint spatial and scale attention network for multi-view facial expression recognition.

TSingNet: Scale-aware and context-rich feature learning for traffic sign detection and recognition in the wild.

Pose-adaptive Hierarchical Attention Network for Facial Expression Recognition.