3D-CNNs

Saliency Tubes: Visual Explanations for Spatio-Temporal Convolutions

Saliency Tubes demonstrate the foremost points and regions in both frame level and over time that are found to be the main focus points of a network.