When reading some deep learning papers, which sometimes mentioned that max-pooling layer for downsampling can also be used for decreasing/increasing the number of feature channels(maps). This confused me a lot. It looks to me the max-pooling layer can down sample the size, but should keep the number of original feature maps. I search the usage of max pool in tensorflow, but there is on arguments for change the feature channels. So, could anybody tell me how to decrease the feature channels by max pool layer? for example, max pool changes the number of feature channels in the following picture. enter image description here
Hiroshi Fukui, Robust Pedestrian Attribute Recognition for an Unbalanced Dataset Using Mini-Batch Training with Rarity Rate