Number of feature maps in convolution neural network -
Number of feature maps in convolution neural network -
i've read articles http://www.codeproject.com/articles/143059/neural-network-for-recognition-of-handwritten-di , when turn one: layer #0: grayness scale image of handwritten character in mnist database padded 29x29 pixel. there 29x29= 841 neurons in input layer. layer #1: convolutional layer 6 (6) feature maps. there 13x13x6 = 1014 neurons, (5x5+1)x6 = 156 weights, , 1014x26 = 26364 connections layer #1 previous layer.
how can six(6) feature maps convolution on image ? think 1 feature map. or wrong ?
you 6 feature maps convolving 6 different kernel on same image.
neural-network convolution
Comments
Post a Comment