Pooling the Convolutional Layers in Deep ConvNets for Video Action Recognition | IEEE Journals & Magazine | IEEE Xplore