DeepLumina: A Method Based on Deep Features and Luminance Information for Color Texture Classification

Table 2

CNN parameters used for feature map generation from the Pretrained Models

CNN parameters	ResNet50	AlexNet	Inceptionv3	DenseNet201
Convlayer name	conv1	conv1	conv2d_1	conv1 | conv
No.of layers	177	25	316	709
Input image size	[224,224,3]	[227,227,3]	[299,299,3]	[224,224,3]
Filter size	7 × 7	11 × 11	3 × 3	7 × 7
No of filters	64	96	32	64
Stride	[2,2]	[4,4]	[2,2]	[2,2]
Dilation factor	[1,1]	[1,1]	[1,1]	[1,1]
Padding size	[3,3,3,3]	[0,0,0,0]	[0,0,0,0]	[3,3,3,3]
Weight initialization	Glorot	Glorot	Glorot	Glorot
Feature layer	fc1000	fc8	Predictions	fc1000