diff --git a/models/CNN_ResNet_21/WiAU_v2.0.prototxt b/models/CNN_ResNet_21/WiAU_v2.0.prototxt
new file mode 100644
index 000000000..5b5e5fdc1
--- /dev/null
+++ b/models/CNN_ResNet_21/WiAU_v2.0.prototxt
@@ -0,0 +1,2528 @@
+name: "WiAU_v2"
+layer {
+  name: "data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    scale: 0.02
+  }
+  data_param {
+    source: "WiAU_v2/train"
+    batch_size: 50
+    backend: LMDB
+  }
+}
+layer {
+  name: "data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    scale: 0.02
+  }
+  data_param {
+    source: "WiAU_v2/test"
+    batch_size: 50
+    backend: LMDB
+  }
+}
+
+layer {
+  name: "conv1"   
+  type: "Convolution"  
+  bottom: "data"   
+  top: "conv1"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 7
+    stride: 2
+    pad: 2
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu1"
+	type:"ReLU"
+	bottom:"conv1"
+	top:"conv1"
+}
+
+layer{
+	name:"pool1"
+	type:"Pooling"
+	bottom:"conv1"
+	top:"pool1"
+	pooling_param{
+		pool:MAX
+		kernel_h: 3
+		kernel_w: 3
+		stride_h: 2
+		stride_w: 2
+		pad: 0
+	}
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+layer {
+  name: "conv_a1_in"
+  type: "Convolution"
+  bottom: "conv2"
+  top: "conv_a1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a1_in"
+	type:"ReLU"
+	bottom:"conv_a1_in"
+	top:"conv_a1_in"
+}
+layer {
+  name: "conv_a1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_a1_in"
+  top: "conv_a1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_a1_bottleneck"
+	top:"conv_a1_bottleneck"
+}
+layer {
+  name: "conv_a1_out"
+  type: "Convolution"
+  bottom: "conv_a1_bottleneck"
+  top: "conv_a1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a1_out"
+	type:"ReLU"
+	bottom:"conv_a1_out"
+	top:"conv_a1_out"
+}
+layer {
+  name: "conv_a1_sum"
+  type: "Eltwise"
+  bottom: "conv_a1_out"
+  bottom: "conv2"
+  top: "conv_a1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_a2_in"
+  type: "Convolution"
+  bottom: "conv_a1_sum"
+  top: "conv_a2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a2_in"
+	type:"ReLU"
+	bottom:"conv_a2_in"
+	top:"conv_a2_in"
+}
+
+layer {
+  name: "conv_a2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_a2_in"
+  top: "conv_a2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_a2_bottleneck"
+	top:"conv_a2_bottleneck"
+}
+layer {
+  name: "conv_a2_out"
+  type: "Convolution"
+  bottom: "conv_a2_bottleneck"
+  top: "conv_a2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a2_out"
+	type:"ReLU"
+	bottom:"conv_a2_out"
+	top:"conv_a2_out"
+}
+layer {
+  name: "conv_a2_sum"
+  type: "Eltwise"
+  bottom: "conv_a2_out"
+  bottom: "conv_a1_sum"
+  top: "conv_a2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_a3_in"
+  type: "Convolution"
+  bottom: "conv_a2_sum"
+  top: "conv_a3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a3_in"
+	type:"ReLU"
+	bottom:"conv_a3_in"
+	top:"conv_a3_in"
+}
+layer {
+  name: "conv_a3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_a3_in"
+  top: "conv_a3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_a3_bottleneck"
+	top:"conv_a3_bottleneck"
+}
+layer {
+  name: "conv_a3_out"
+  type: "Convolution"
+  bottom: "conv_a3_bottleneck"
+  top: "conv_a3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_a3_out"
+	type:"ReLU"
+	bottom:"conv_a3_out"
+	top:"conv_a3_out"
+}
+layer {
+  name: "conv_a3_sum"
+  type: "Eltwise"
+  bottom: "conv_a3_out"
+  bottom: "conv_a2_sum"
+  top: "conv_a3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_a3_upscale"
+  type: "Convolution"
+  bottom: "conv_a3_sum"
+  top: "conv_a3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_b1_in"
+  type: "Convolution"
+  bottom: "conv_a3_upscale"
+  top: "conv_b1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b1_in"
+	type:"ReLU"
+	bottom:"conv_b1_in"
+	top:"conv_b1_in"
+}
+layer {
+  name: "conv_b1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_b1_in"
+  top: "conv_b1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_b1_bottleneck"
+	top:"conv_b1_bottleneck"
+}
+layer {
+  name: "conv_b1_out"
+  type: "Convolution"
+  bottom: "conv_b1_bottleneck"
+  top: "conv_b1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b1_out"
+	type:"ReLU"
+	bottom:"conv_b1_out"
+	top:"conv_b1_out"
+}
+layer {
+  name: "conv_b1_sum"
+  type: "Eltwise"
+  bottom: "conv_b1_out"
+  bottom: "conv_a3_sum"
+  top: "conv_b1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_b2_in"
+  type: "Convolution"
+  bottom: "conv_b1_sum"
+  top: "conv_b2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b2_in"
+	type:"ReLU"
+	bottom:"conv_b2_in"
+	top:"conv_b2_in"
+}
+
+layer {
+  name: "conv_b2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_b2_in"
+  top: "conv_b2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_b2_bottleneck"
+	top:"conv_b2_bottleneck"
+}
+layer {
+  name: "conv_b2_out"
+  type: "Convolution"
+  bottom: "conv_b2_bottleneck"
+  top: "conv_b2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b2_out"
+	type:"ReLU"
+	bottom:"conv_b2_out"
+	top:"conv_b2_out"
+}
+layer {
+  name: "conv_b2_sum"
+  type: "Eltwise"
+  bottom: "conv_b2_out"
+  bottom: "conv_b1_sum"
+  top: "conv_b2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_b3_in"
+  type: "Convolution"
+  bottom: "conv_b2_sum"
+  top: "conv_b3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b3_in"
+	type:"ReLU"
+	bottom:"conv_b3_in"
+	top:"conv_b3_in"
+}
+layer {
+  name: "conv_b3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_b3_in"
+  top: "conv_b3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 32
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_b3_bottleneck"
+	top:"conv_b3_bottleneck"
+}
+layer {
+  name: "conv_b3_out"
+  type: "Convolution"
+  bottom: "conv_b3_bottleneck"
+  top: "conv_b3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_b3_out"
+	type:"ReLU"
+	bottom:"conv_b3_out"
+	top:"conv_b3_out"
+}
+layer {
+  name: "conv_b3_sum"
+  type: "Eltwise"
+  bottom: "conv_b3_out"
+  bottom: "conv_b2_sum"
+  top: "conv_b3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_b3_upscale"
+  type: "Convolution"
+  bottom: "conv_b3_sum"
+  top: "conv_b3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_c1_in"
+  type: "Convolution"
+  bottom: "conv_b3_upscale"
+  top: "conv_c1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c1_in"
+	type:"ReLU"
+	bottom:"conv_c1_in"
+	top:"conv_c1_in"
+}
+layer {
+  name: "conv_c1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_c1_in"
+  top: "conv_c1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_c1_bottleneck"
+	top:"conv_c1_bottleneck"
+}
+layer {
+  name: "conv_c1_out"
+  type: "Convolution"
+  bottom: "conv_c1_bottleneck"
+  top: "conv_c1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c1_out"
+	type:"ReLU"
+	bottom:"conv_c1_out"
+	top:"conv_c1_out"
+}
+layer {
+  name: "conv_c1_sum"
+  type: "Eltwise"
+  bottom: "conv_c1_out"
+  bottom: "conv_b3_sum"
+  top: "conv_c1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_c2_in"
+  type: "Convolution"
+  bottom: "conv_c1_sum"
+  top: "conv_c2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c2_in"
+	type:"ReLU"
+	bottom:"conv_c2_in"
+	top:"conv_c2_in"
+}
+
+layer {
+  name: "conv_c2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_c2_in"
+  top: "conv_c2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_c2_bottleneck"
+	top:"conv_c2_bottleneck"
+}
+layer {
+  name: "conv_c2_out"
+  type: "Convolution"
+  bottom: "conv_c2_bottleneck"
+  top: "conv_c2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c2_out"
+	type:"ReLU"
+	bottom:"conv_c2_out"
+	top:"conv_c2_out"
+}
+layer {
+  name: "conv_c2_sum"
+  type: "Eltwise"
+  bottom: "conv_c2_out"
+  bottom: "conv_c1_sum"
+  top: "conv_c2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_c3_in"
+  type: "Convolution"
+  bottom: "conv_c2_sum"
+  top: "conv_c3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c3_in"
+	type:"ReLU"
+	bottom:"conv_c3_in"
+	top:"conv_c3_in"
+}
+layer {
+  name: "conv_c3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_c3_in"
+  top: "conv_c3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_c3_bottleneck"
+	top:"conv_c3_bottleneck"
+}
+layer {
+  name: "conv_c3_out"
+  type: "Convolution"
+  bottom: "conv_c3_bottleneck"
+  top: "conv_c3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_c3_out"
+	type:"ReLU"
+	bottom:"conv_c3_out"
+	top:"conv_c3_out"
+}
+layer {
+  name: "conv_c3_sum"
+  type: "Eltwise"
+  bottom: "conv_c3_out"
+  bottom: "conv_c2_sum"
+  top: "conv_c3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_c3_upscale"
+  type: "Convolution"
+  bottom: "conv_c3_sum"
+  top: "conv_c3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_d1_in"
+  type: "Convolution"
+  bottom: "conv_c3_upscale"
+  top: "conv_d1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d1_in"
+	type:"ReLU"
+	bottom:"conv_d1_in"
+	top:"conv_d1_in"
+}
+layer {
+  name: "conv_d1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_d1_in"
+  top: "conv_d1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_d1_bottleneck"
+	top:"conv_d1_bottleneck"
+}
+layer {
+  name: "conv_d1_out"
+  type: "Convolution"
+  bottom: "conv_d1_bottleneck"
+  top: "conv_d1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d1_out"
+	type:"ReLU"
+	bottom:"conv_d1_out"
+	top:"conv_d1_out"
+}
+layer {
+  name: "conv_d1_sum"
+  type: "Eltwise"
+  bottom: "conv_d1_out"
+  bottom: "conv_c3_sum"
+  top: "conv_d1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_d2_in"
+  type: "Convolution"
+  bottom: "conv_d1_sum"
+  top: "conv_d2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d2_in"
+	type:"ReLU"
+	bottom:"conv_d2_in"
+	top:"conv_d2_in"
+}
+
+layer {
+  name: "conv_d2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_d2_in"
+  top: "conv_d2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_d2_bottleneck"
+	top:"conv_d2_bottleneck"
+}
+layer {
+  name: "conv_d2_out"
+  type: "Convolution"
+  bottom: "conv_d2_bottleneck"
+  top: "conv_d2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d2_out"
+	type:"ReLU"
+	bottom:"conv_d2_out"
+	top:"conv_d2_out"
+}
+layer {
+  name: "conv_d2_sum"
+  type: "Eltwise"
+  bottom: "conv_d2_out"
+  bottom: "conv_d1_sum"
+  top: "conv_d2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_d3_in"
+  type: "Convolution"
+  bottom: "conv_d2_sum"
+  top: "conv_d3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d3_in"
+	type:"ReLU"
+	bottom:"conv_d3_in"
+	top:"conv_d3_in"
+}
+layer {
+  name: "conv_d3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_d3_in"
+  top: "conv_d3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 64
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_d3_bottleneck"
+	top:"conv_d3_bottleneck"
+}
+layer {
+  name: "conv_d3_out"
+  type: "Convolution"
+  bottom: "conv_d3_bottleneck"
+  top: "conv_d3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_d3_out"
+	type:"ReLU"
+	bottom:"conv_d3_out"
+	top:"conv_d3_out"
+}
+layer {
+  name: "conv_d3_sum"
+  type: "Eltwise"
+  bottom: "conv_d3_out"
+  bottom: "conv_d2_sum"
+  top: "conv_d3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_d3_upscale"
+  type: "Convolution"
+  bottom: "conv_d3_sum"
+  top: "conv_d3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_e1_in"
+  type: "Convolution"
+  bottom: "conv_d3_upscale"
+  top: "conv_e1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e1_in"
+	type:"ReLU"
+	bottom:"conv_e1_in"
+	top:"conv_e1_in"
+}
+layer {
+  name: "conv_e1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_e1_in"
+  top: "conv_e1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_e1_bottleneck"
+	top:"conv_e1_bottleneck"
+}
+layer {
+  name: "conv_e1_out"
+  type: "Convolution"
+  bottom: "conv_e1_bottleneck"
+  top: "conv_e1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e1_out"
+	type:"ReLU"
+	bottom:"conv_e1_out"
+	top:"conv_e1_out"
+}
+layer {
+  name: "conv_e1_sum"
+  type: "Eltwise"
+  bottom: "conv_e1_out"
+  bottom: "conv_d3_sum"
+  top: "conv_e1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_e2_in"
+  type: "Convolution"
+  bottom: "conv_e1_sum"
+  top: "conv_e2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e2_in"
+	type:"ReLU"
+	bottom:"conv_e2_in"
+	top:"conv_e2_in"
+}
+
+layer {
+  name: "conv_e2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_e2_in"
+  top: "conv_e2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_e2_bottleneck"
+	top:"conv_e2_bottleneck"
+}
+layer {
+  name: "conv_e2_out"
+  type: "Convolution"
+  bottom: "conv_e2_bottleneck"
+  top: "conv_e2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e2_out"
+	type:"ReLU"
+	bottom:"conv_e2_out"
+	top:"conv_e2_out"
+}
+layer {
+  name: "conv_e2_sum"
+  type: "Eltwise"
+  bottom: "conv_e2_out"
+  bottom: "conv_e1_sum"
+  top: "conv_e2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_e3_in"
+  type: "Convolution"
+  bottom: "conv_e2_sum"
+  top: "conv_e3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e3_in"
+	type:"ReLU"
+	bottom:"conv_e3_in"
+	top:"conv_e3_in"
+}
+layer {
+  name: "conv_e3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_e3_in"
+  top: "conv_e3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_e3_bottleneck"
+	top:"conv_e3_bottleneck"
+}
+layer {
+  name: "conv_e3_out"
+  type: "Convolution"
+  bottom: "conv_e3_bottleneck"
+  top: "conv_e3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_e3_out"
+	type:"ReLU"
+	bottom:"conv_e3_out"
+	top:"conv_e3_out"
+}
+layer {
+  name: "conv_e3_sum"
+  type: "Eltwise"
+  bottom: "conv_e3_out"
+  bottom: "conv_e2_sum"
+  top: "conv_e3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_e3_upscale"
+  type: "Convolution"
+  bottom: "conv_e3_sum"
+  top: "conv_e3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_f1_in"
+  type: "Convolution"
+  bottom: "conv_e3_upscale"
+  top: "conv_f1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f1_in"
+	type:"ReLU"
+	bottom:"conv_f1_in"
+	top:"conv_f1_in"
+}
+layer {
+  name: "conv_f1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_f1_in"
+  top: "conv_f1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_f1_bottleneck"
+	top:"conv_f1_bottleneck"
+}
+layer {
+  name: "conv_f1_out"
+  type: "Convolution"
+  bottom: "conv_f1_bottleneck"
+  top: "conv_f1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f1_out"
+	type:"ReLU"
+	bottom:"conv_f1_out"
+	top:"conv_f1_out"
+}
+layer {
+  name: "conv_f1_sum"
+  type: "Eltwise"
+  bottom: "conv_f1_out"
+  bottom: "conv_e3_sum"
+  top: "conv_f1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_f2_in"
+  type: "Convolution"
+  bottom: "conv_f1_sum"
+  top: "conv_f2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f2_in"
+	type:"ReLU"
+	bottom:"conv_f2_in"
+	top:"conv_f2_in"
+}
+
+layer {
+  name: "conv_f2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_f2_in"
+  top: "conv_f2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_f2_bottleneck"
+	top:"conv_f2_bottleneck"
+}
+layer {
+  name: "conv_f2_out"
+  type: "Convolution"
+  bottom: "conv_f2_bottleneck"
+  top: "conv_f2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f2_out"
+	type:"ReLU"
+	bottom:"conv_f2_out"
+	top:"conv_f2_out"
+}
+layer {
+  name: "conv_f2_sum"
+  type: "Eltwise"
+  bottom: "conv_f2_out"
+  bottom: "conv_f1_sum"
+  top: "conv_f2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_f3_in"
+  type: "Convolution"
+  bottom: "conv_f2_sum"
+  top: "conv_f3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f3_in"
+	type:"ReLU"
+	bottom:"conv_f3_in"
+	top:"conv_f3_in"
+}
+layer {
+  name: "conv_f3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_f3_in"
+  top: "conv_f3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 128
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_f3_bottleneck"
+	top:"conv_f3_bottleneck"
+}
+layer {
+  name: "conv_f3_out"
+  type: "Convolution"
+  bottom: "conv_f3_bottleneck"
+  top: "conv_f3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_f3_out"
+	type:"ReLU"
+	bottom:"conv_f3_out"
+	top:"conv_f3_out"
+}
+layer {
+  name: "conv_f3_sum"
+  type: "Eltwise"
+  bottom: "conv_f3_out"
+  bottom: "conv_f2_sum"
+  top: "conv_f3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_f3_upscale"
+  type: "Convolution"
+  bottom: "conv_f3_sum"
+  top: "conv_f3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 512
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer {
+  name: "conv_g1_in"
+  type: "Convolution"
+  bottom: "conv_f3_upscale"
+  top: "conv_g1_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g1_in"
+	type:"ReLU"
+	bottom:"conv_g1_in"
+	top:"conv_g1_in"
+}
+layer {
+  name: "conv_g1_bottleneck"
+  type: "Convolution"
+  bottom: "conv_g1_in"
+  top: "conv_g1_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g1_bottleneck"
+	type:"ReLU"
+	bottom:"conv_g1_bottleneck"
+	top:"conv_g1_bottleneck"
+}
+layer {
+  name: "conv_g1_out"
+  type: "Convolution"
+  bottom: "conv_g1_bottleneck"
+  top: "conv_g1_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 1024
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g1_out"
+	type:"ReLU"
+	bottom:"conv_g1_out"
+	top:"conv_g1_out"
+}
+layer {
+  name: "conv_g1_sum"
+  type: "Eltwise"
+  bottom: "conv_g1_out"
+  bottom: "conv_f3_sum"
+  top: "conv_g1_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+
+layer {
+  name: "conv_g2_in"
+  type: "Convolution"
+  bottom: "conv_g1_sum"
+  top: "conv_g2_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g2_in"
+	type:"ReLU"
+	bottom:"conv_g2_in"
+	top:"conv_g2_in"
+}
+
+layer {
+  name: "conv_g2_bottleneck"
+  type: "Convolution"
+  bottom: "conv_g2_in"
+  top: "conv_g2_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g2_bottleneck"
+	type:"ReLU"
+	bottom:"conv_g2_bottleneck"
+	top:"conv_g2_bottleneck"
+}
+layer {
+  name: "conv_g2_out"
+  type: "Convolution"
+  bottom: "conv_g2_bottleneck"
+  top: "conv_g2_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 1024
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g2_out"
+	type:"ReLU"
+	bottom:"conv_g2_out"
+	top:"conv_g2_out"
+}
+layer {
+  name: "conv_g2_sum"
+  type: "Eltwise"
+  bottom: "conv_g2_out"
+  bottom: "conv_g1_sum"
+  top: "conv_g2_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_g3_in"
+  type: "Convolution"
+  bottom: "conv_g2_sum"
+  top: "conv_g3_in"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g3_in"
+	type:"ReLU"
+	bottom:"conv_g3_in"
+	top:"conv_g3_in"
+}
+layer {
+  name: "conv_g3_bottleneck"
+  type: "Convolution"
+  bottom: "conv_g3_in"
+  top: "conv_g3_bottleneck"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 256
+    kernel_size: 3
+    stride: 1
+	pad: 1
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g3_bottleneck"
+	type:"ReLU"
+	bottom:"conv_g3_bottleneck"
+	top:"conv_g3_bottleneck"
+}
+layer {
+  name: "conv_g3_out"
+  type: "Convolution"
+  bottom: "conv_g3_bottleneck"
+  top: "conv_g3_out"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 1024
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+layer{
+	name:"relu_g3_out"
+	type:"ReLU"
+	bottom:"conv_g3_out"
+	top:"conv_g3_out"
+}
+layer {
+  name: "conv_g3_sum"
+  type: "Eltwise"
+  bottom: "conv_g3_out"
+  bottom: "conv_g2_sum"
+  top: "conv_g3_sum"
+  eltwise_param {
+    operation: SUM
+  }
+}
+
+layer {
+  name: "conv_g3_upscale"
+  type: "Convolution"
+  bottom: "conv_g3_sum"
+  top: "conv_g3_upscale"
+  param {
+    lr_mult: 1   
+  }
+  param {
+    lr_mult: 2
+  }
+  convolution_param {
+    num_output: 1024
+    kernel_size: 1
+    stride: 1
+	pad: 0
+    weight_filler {
+      type: "xavier"
+    }
+    bias_filler {
+      type: "constant"
+    }
+  }
+}
+
+
+
+
+
+
+
+
+
+layer{
+	name:"pool2"
+	type:"Pooling"
+	bottom:"conv_g3_upscale"
+	top:"pool2"
+	pooling_param{
+		pool:AVG
+		kernel_h: 2
+		kernel_w: 2
+		stride_h: 1
+		stride_w: 1
+		pad: 0
+	}
+}
+layer{
+	name:"reshape1"
+	type:"Reshape"
+	bottom:"pool2"
+	top:"reshape1"
+	reshape_param {
+      shape {
+        dim:0
+		dim:0
+		dim:32
+		dim:-1
+      }
+    }
+}
+
+
+
+
+layer {  
+  name: "loss"  
+  type: "SoftmaxWithLoss"  
+  bottom: "reshape1"  
+  bottom: "label"  
+  top: "loss"  
+}  
+layer{
+	name:"cross"
+	type:"SigmoidCrossEntropyLossLayer"
+	bottom:"loss"
+	bottom:"label"
+	top:"cross"
+}
+layer {
+  name: "Accuracy"
+  type: "Accuracy"
+  bottom: "loss"
+  bottom: "label"
+  top: "Accuracy"
+}
diff --git a/models/CNN_ResNet_21/solver.prototxt b/models/CNN_ResNet_21/solver.prototxt
new file mode 100644
index 000000000..0684bfc7b
--- /dev/null
+++ b/models/CNN_ResNet_21/solver.prototxt
@@ -0,0 +1,22 @@
+net: "WiAU_2.0_CNN+ResNet.prototxt"
+
+test_iter: 100 
+
+test_interval: 400
+
+base_lr: 0.01   
+momentum: 0.9   
+weight_decay: 0.0005  
+
+lr_policy: "inv" 
+gamma: 0.0001
+power: 0.75
+
+display: 100
+
+max_iter: 100000
+
+snapshot: 5000
+snapshot_prefix: "WiAU_v2/result"
+
+solver_mode: GPU
diff --git a/python/Transform.py b/python/Transform.py
new file mode 100644
index 000000000..775960e2a
--- /dev/null
+++ b/python/Transform.py
@@ -0,0 +1,247 @@
+import numpy as np
+import matplotlib.pyplot as plt
+from matplotlib import animation
+import os
+import pickle
+import cmath
+import math
+import scipy.stats as scita
+import scipy.signal as signal
+import scipy.io as scio
+from numpy import linalg as la
+from numpy import *
+import lmdb
+import caffe
+
+winLen=30
+
+def db(x):
+#变换到log空间
+	if (x==0):
+		ans=0
+	else:
+		ans=20*(np.log10(x))
+	return ans
+
+def butterworth_II(file,Fc):
+#去噪
+	N  = 2    # Filter order
+	Wn = 2*(np.pi)*Fc # Cutoff frequency
+	B, A = signal.butter(N, Wn, output='ba')
+	ret = signal.filtfilt(B,A,file)
+	return ret
+
+def relative_phase(tmp1,tmp2):
+#计算相对相位
+	tmp=tmp1*np.conjugate(tmp2)
+	tmp_1=(tmp.real)/(abs(tmp))
+	ret=np.arccos(tmp_1)
+
+	return (ret)
+
+
+
+
+def file_data(filename):
+#解析原始数据包
+	l=[]
+	with open(filename, "rb") as f:
+		while 1:
+			try:
+				flag = 1
+				k = pickle.load(f)
+				for i in range(90):
+					if(abs(k[i]) <= 0):
+						flag = 0
+				if flag == 0:
+					continue
+				l.append(k[0:90])
+			except Exception as e:
+				break
+	a = np.array(l).T
+	return(a)
+
+def csi_amplitude(file):
+	[row,col]=file.shape
+	newFile = np.zeros((row,col))
+	for i in range(row):
+		for j in range(col):
+			newFile[i,j]=db(abs(file[i,j]))
+			#print (file[i,j])
+	ret=np.array(newFile)
+	return (ret.real)
+
+def csi_relative_phase(file):
+#计算数据的相对相位
+	file=file.reshape(3,30,-1)
+	[row,col,other]=file.shape
+	csi_ant1=file[0]
+	csi_ant2=file[1]
+	csi_ant3=file[2]
+	rephase1_2=[]
+	rephase1_3=[]
+	rephase3_2=[]
+	rephase_all=[]
+
+	for i in range(col):
+		tmp1_2=relative_phase(csi_ant1[i],csi_ant2[i])
+
+		tmp1_3=relative_phase(csi_ant1[i],csi_ant3[i])
+		tmp3_2=relative_phase(csi_ant3[i],csi_ant2[i])
+		rephase1_2.append(tmp1_2)
+		rephase1_3.append(tmp1_3)
+		rephase3_2.append(tmp3_2)
+
+	for j in range(30):
+		rephase_all.append(rephase1_2[j])
+	for j in range(30):
+		rephase_all.append(rephase1_3[j])
+	for j in range(30):
+		rephase_all.append(rephase3_2[j])
+
+	ret=np.array(rephase_all)
+	return ret.real
+
+def get_characters(matrix, num, label):
+#计算出相应特征值
+	max = []
+	min = []
+	mean = []
+	skewness = []
+	kurtosis = []
+	std = []
+	i = 1
+	col = matrix.shape[1]
+	chunk = int(col / num)
+	while (i) * chunk <= col and i <= num:
+		tmp = matrix[:, chunk * (i-1):chunk * (i)]
+		i = i + 1
+		cnt = 0
+		max_t = []
+		min_t = []
+		mean_t = []
+		skewness_t = []
+		kurtosis_t = []
+		std_t = []
+		while cnt < 90:
+			t = tmp[cnt:]
+			max_t.append(np.max(t))
+			min_t.append(np.min(t))
+			mean_t.append(np.mean(t))
+			skewness_t.append(np.mean(scita.skew(t,axis=1, bias=True)))
+			kurtosis_t.append(np.mean(scita.kurtosis(t,axis=1, bias=True)))
+			std_t.append(np.std(t))
+			cnt = cnt + 1
+		max.append(max_t)
+		min.append(min_t)
+		mean.append(mean_t)
+		skewness.append(skewness_t)
+		kurtosis.append(kurtosis_t)
+		std.append(std_t)
+	l = []
+	for i in range(0, 540):
+		l.append(label)
+	max = np.array(max).T
+	min = np.array(min).T
+	mean = np.array(mean).T
+	skewness = np.array(skewness).T
+	kurtosis = np.array(kurtosis).T
+	std = np.array(std).T
+	result = np.append(max, min, axis=0)
+	result = np.append(result, mean, axis=0)
+	result = np.append(result, skewness, axis=0)
+	result = np.append(result, kurtosis, axis=0)
+	result = np.append(result, std, axis=0)
+	return result, l
+
+def classification(input):
+#根据标签list转化为标签矩阵（one-hot编码）
+	labelOrder=set()
+	labelList=[]
+	i=0
+	for label in input:
+		label=label.split('_')[0]
+		labelList.append(label)
+		labelOrder.add(label)
+	labelOrder=list(labelOrder)
+	labelCol=len(labelOrder)
+	labelMatrix = []
+	for label in labelList:
+		l = [0] * labelCol
+		labelNum = labelOrder.index(label)
+		l[labelNum] = 1
+		labelMatrix.append(l)
+	labelMatrix = np.array(labelMatrix, dtype=object)
+	print(labelMatrix.shape)
+	return labelMatrix,labelOrder
+
+
+		
+			
+	
+
+def train_data(path,num):
+#输入为路径和特征值数量
+	files= os.listdir(path)
+	labelList=[]
+	featureList=[]
+	for file in files:
+		if file == 'over':
+			continue
+		for x in os.listdir(path + '/' + file):
+			print(x)
+			csiData=file_data(str(path) + '/' + file + '/' + x)
+			csiAmplitude=csi_amplitude(csiData)
+			csiAmplitude=butterworth_II(csiAmplitude,0.03)
+			csiFeature, labels=get_characters(csiAmplitude,num,file)
+			labelList.extend(labels)
+			featureList.extend(csiFeature)
+	labelTmp=np.array(labelList)
+	featureMatrix=np.array(featureList)
+	labelMatrix,labelOrder=classification(labelTmp)
+	return featureMatrix,labelMatrix,labelOrder
+
+
+def test_input(path,num):
+	files = os.listdir(path)
+	print(files)
+	featureList = []
+	for file in files:
+		csiData = file_data(str(path) + '/' + file)
+		csiAmplitude = csi_amplitude(csiData)
+		csiAmplitude = butterworth_II(csiAmplitude, 0.03)
+		csiFeature, labels = get_characters(csiAmplitude, num, file)
+		featureList.extend(csiFeature)
+	featureMatrix = np.array(featureList)
+	return featureMatrix
+
+def turn_to_lmdb():#将csi数据转换为lmdb格式主文件
+    data,label,labelOrder1=train_data(rootpath + '\\datatrain', 784)
+    N = 1000
+
+# Let's pretend this is interesting data
+    X = data
+    y = label
+
+# We need to prepare the database for the size. We'll set it 10 times
+# greater than what we theoretically need. There is little drawback to
+# setting this too big. If you still run into problem after raising
+# this, you might want to try saving fewer entries in a single
+# transaction.
+    map_size = X.nbytes * 10
+    env = lmdb.open('mylmdb', map_size=map_size)
+
+    with env.begin(write=True) as txn:
+    # txn is a Transaction object
+        for i in range(N):
+            datum = caffe.proto.caffe_pb2.Datum()#处理好这五个变量即可
+            datum.channels = X.shape[1]
+            datum.height = X.shape[2]
+            datum.width = X.shape[3]
+            datum.data = X[i].tobytes()  # or .tostring() if numpy < 1.9
+            datum.label = int(y[i])
+            str_id = '{:08}'.format(i)
+
+        # The encode is only essential in Python 3
+            txn.put(str_id.encode('ascii'), datum.SerializeToString())
+    env.close