better model and config

model_pixel_up.py

@@ -10,7 +10,16 @@ def build_model(x, scale, training, reuse):
     for i in range(6):
         x = util.crop_by_pixel(x, 1) + conv(x, hidden_size, bottleneck_size, training, 'lr_conv'+str(i), reuse)
     x = util.lrelu(x)
-    x = tf.image.resize_nearest_neighbor(x, tf.shape(x)[1:3] * scale) + tf.layers.conv2d_transpose(x, hidden_size, scale, strides=scale, activation=None, name='up', reuse=reuse)
+    if (scale == 4):
+        scale = 2
+        x = tf.layers.conv2d_transpose(x, hidden_size, scale, strides=scale, activation=None, name='up1', reuse=reuse)
+        x = util.crop_by_pixel(x, 1) + conv(x, hidden_size, bottleneck_size, training, 'up_conv', reuse)
+        x = util.lrelu(x)
+        hidden_size = 64
+        x = tf.layers.conv2d_transpose(x, hidden_size, scale, strides=scale, activation=None, name='up2', reuse=reuse)
+    else:
+        hidden_size = 64
+        x = tf.layers.conv2d_transpose(x, hidden_size, scale, strides=scale, activation=None, name='up', reuse=reuse)
     for i in range(4):
         x = util.crop_by_pixel(x, 1) + conv(x, hidden_size, bottleneck_size, training, 'hr_conv'+str(i), reuse)
     x = util.lrelu(x)

run.sh

@@ -24,10 +24,10 @@ set -x
 
 EXPR_NAME="try"
 TRAIN_DIR="tmp"
-MODEL_NAME="model_resnet_up"
+MODEL_NAME="model_pixel_up"
 DATA_NAME="data_residual"
-HR_FLIST="flist/hr.flist"
-LR_FLIST="flist/lrX2.flist"
+HR_FLIST="flist/hr_tv.flist"
+LR_FLIST="flist/lrX2_bicubic_tv.flist"
 SCALE=2
 LEARNING_RATE=0.001
 
@@ -49,10 +49,10 @@ ARGS="--data_name=$DATA_NAME --hr_flist=$HR_FLIST --lr_flist=$LR_FLIST --model_n
 
 iter=0
 rate=$LEARNING_RATE
-for i in `seq 1 8`;
+for i in `seq 1 16`;
 do
     python $SCRIPT $ARGS --model_file_in=$MODEL_FILE-$iter --model_file_out=$MODEL_FILE-$((iter+1)) --learning_rate=$rate
     iter=$((iter+1))
-    rate=$(echo "$rate" | awk '{print $1*0.5}')
+    rate=$(echo "$rate" | awk '{print $1*0.618}')
     echo "Iteration $iter Finished"
 done