Open bingyizhao opened 3 months ago
Same problem With this numbers layers are the same with the article conv4 = backbone.layers[758].output conv3 = backbone.layers[444].output conv2 = backbone.layers[177].output conv1 = backbone.layers[73].output
同样的问题 对于这个数字,层与文章相同 conv4 = backbone.layers[758].output conv3 = backbone.layers[444].output conv2 = backbone.layers[177].output conv1 = backbone.layers[73].output
Is the number of layers in this number, e.g. con4=layer[758], the layer information of the optimal structure obtained by training based on remote sensing data?
Why can't the layers of the EfficientNetB7 network structure correspond to the block names? For example, layer [30] corresponds to 1c.
backbone.summary(): layer_30_block1c_dwconv (D (None, 128, 128, 32) 288 ['layer_29_block1b_add[0][0]'] epthwiseConv2D)