[foundation of deep learning] how to understand the channel in convolutional neural network

What is a channel channel

【CNN】 Understand the channels in convolutional neural networks channel

First , yes tensorflow Given in , For input samples channels The meaning of . General RGB picture ,channels The number is 3 （ red 、 green 、 blue ）; and monochrome picture ,channels The number is 1 .

secondly ,mxnet Mentioned in , commonly channels The meaning is , The number of convolution kernels in each convolution layer .

give an example ： Suppose the existing one is 6×6×3 Sample pictures of , Use 3×3×3 Convolution kernel （filter） Convolution operation . At this point, enter the name of the picture channels by 3 , And in the convolution kernel in_channels And Of data requiring convolution channels Agreement （ Here is the sample picture , by 3）.

Next , Convolution operation , In convolution kernel 27 A number is multiplied by the corresponding sample , And then sum it up , Get the first result . In turn , The resulting 4×4 Result .

After the above steps are completed , Because there is only one convolution kernel , So the final result is 4×4×1 , out_channels by 1 .

in application , Will use multiple convolution kernels . If you add another convolution kernel here , Will get 4×4×2 Result .

summary

• Of the original input image sample channels , Depending on the picture type , such as RGB;
• Output after convolution operation out_channels , Depending on the number of convolution kernels . At this time out_channels It will also be used as the convolution kernel for the next convolution in_channels;
• In convolution kernel in_channels , It's the result of the last convolution out_channels , If you're doing convolution for the first time , Namely 1 In the sample picture channels .

This blog post was first published on the personal blog website ：https://www.mahaofei.com/, Welcome to .