model_infer.py

import tensorflow as tf
from networks import Network
from utils.image_reader import _infer_preprocess
from utils.visualize import decode_labels

class ICNet(Network):
    def __init__(self, cfg, mode, image_reader=None):
        self.cfg = cfg
        self.mode = mode

        if mode == 'train':
            self.images, self.labels = image_reader.next_image, image_reader.next_label    
        
            super().__init__(inputs={'data': self.images}, cfg=self.cfg)

        elif mode == 'eval':
            self.images, self.labels = image_reader.next_image, image_reader.next_label    
        
            super().__init__(inputs={'data': self.images}, cfg=self.cfg)
            
            self.output = self.get_output_node()

        elif mode == 'inference':
            # Create placeholder and pre-process here.
            self.img_placeholder = tf.placeholder(dtype=tf.float32, shape=cfg.INFER_SIZE, name="input")
            self.images, self.o_shape, self.n_shape = _infer_preprocess(self.img_placeholder)
            
            super(ICNet, self).__init__(inputs={'data': self.images}, cfg=self.cfg)

            self.output = self.get_output_node()

    def get_output_node(self):
        if self.mode == 'inference':
            # Get logits from final layer
            logits = self.layers['conv6_cls']

            # Upscale the logits and decode prediction to get final result.
            logits_up = tf.image.resize_bilinear(logits, size=self.n_shape, align_corners=True)
            #logits_up = tf.image.crop_to_bounding_box(logits_up, 0, 0, self.o_shape[0], self.o_shape[1])

            #output_classes = tf.argmax(logits_up, axis=3)
            #output = decode_labels(output_classes, self.o_shape, self.cfg.param['num_classes'])
            ##output = tf.reshape(output_classes, (1, self.o_shape[0], self.o_shape[1], 3))
            output = logits_up

        elif self.mode == 'eval':
            logits = self.layers['conv6_cls']

            logits_up = tf.image.resize_bilinear(logits, size=tf.shape(self.labels)[1:3], align_corners=True)
            output = tf.argmax(logits_up, axis=3)
            output = tf.expand_dims(output, axis=3)

        return output

    def predict(self, image):
        return self.sess.run(self.output, feed_dict={self.img_placeholder: image})

    def setup(self):
        (self.feed('data')
             .interp(s_factor=0.5, name='data_sub2')
             .conv(3, 3, 32, 2, 2, biased=True, padding='SAME', relu=True, name='conv1_1_3x3_s2')
             .conv(3, 3, 32, 1, 1, biased=True, padding='SAME', relu=True, name='conv1_2_3x3')
             .conv(3, 3, 64, 1, 1, biased=True, padding='SAME', relu=True, name='conv1_3_3x3')
             .zero_padding(paddings=1, name='padding0')
             .max_pool(3, 3, 2, 2, name='pool1_3x3_s2')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv2_1_1x1_proj'))

        (self.feed('pool1_3x3_s2')
             .conv(1, 1, 32, 1, 1, biased=True, relu=True, name='conv2_1_1x1_reduce')
             .zero_padding(paddings=1, name='padding1')
             .conv(3, 3, 32, 1, 1, biased=True, relu=True, name='conv2_1_3x3')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv2_1_1x1_increase'))

        (self.feed('conv2_1_1x1_proj',
                   'conv2_1_1x1_increase')
             .add(name='conv2_1')
             .relu(name='conv2_1/relu')
             .conv(1, 1, 32, 1, 1, biased=True, relu=True, name='conv2_2_1x1_reduce')
             .zero_padding(paddings=1, name='padding2')
             .conv(3, 3, 32, 1, 1, biased=True, relu=True, name='conv2_2_3x3')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv2_2_1x1_increase'))

        (self.feed('conv2_1/relu',
                   'conv2_2_1x1_increase')
             .add(name='conv2_2')
             .relu(name='conv2_2/relu')
             .conv(1, 1, 32, 1, 1, biased=True, relu=True, name='conv2_3_1x1_reduce')
             .zero_padding(paddings=1, name='padding3')
             .conv(3, 3, 32, 1, 1, biased=True, relu=True, name='conv2_3_3x3')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv2_3_1x1_increase'))

        (self.feed('conv2_2/relu',
                   'conv2_3_1x1_increase')
             .add(name='conv2_3')
             .relu(name='conv2_3/relu')
             .conv(1, 1, 256, 2, 2, biased=True, relu=False, name='conv3_1_1x1_proj'))

        (self.feed('conv2_3/relu')
             .conv(1, 1, 64, 2, 2, biased=True, relu=True, name='conv3_1_1x1_reduce')
             .zero_padding(paddings=1, name='padding4')
             .conv(3, 3, 64, 1, 1, biased=True, relu=True, name='conv3_1_3x3')
             .conv(1, 1, 256, 1, 1, biased=True, relu=False, name='conv3_1_1x1_increase'))

        (self.feed('conv3_1_1x1_proj',
                   'conv3_1_1x1_increase')
             .add(name='conv3_1')
             .relu(name='conv3_1/relu')
             .interp(s_factor=0.5, name='conv3_1_sub4')
             .conv(1, 1, 64, 1, 1, biased=True, relu=True, name='conv3_2_1x1_reduce')
             .zero_padding(paddings=1, name='padding5')
             .conv(3, 3, 64, 1, 1, biased=True, relu=True, name='conv3_2_3x3')
             .conv(1, 1, 256, 1, 1, biased=True, relu=False, name='conv3_2_1x1_increase'))

        (self.feed('conv3_1_sub4',
                   'conv3_2_1x1_increase')
             .add(name='conv3_2')
             .relu(name='conv3_2/relu')
             .conv(1, 1, 64, 1, 1, biased=True, relu=True, name='conv3_3_1x1_reduce')
             .zero_padding(paddings=1, name='padding6')
             .conv(3, 3, 64, 1, 1, biased=True, relu=True, name='conv3_3_3x3')
             .conv(1, 1, 256, 1, 1, biased=True, relu=False, name='conv3_3_1x1_increase'))

        (self.feed('conv3_2/relu',
                   'conv3_3_1x1_increase')
             .add(name='conv3_3')
             .relu(name='conv3_3/relu')
             .conv(1, 1, 64, 1, 1, biased=True, relu=True, name='conv3_4_1x1_reduce')
             .zero_padding(paddings=1, name='padding7')
             .conv(3, 3, 64, 1, 1, biased=True, relu=True, name='conv3_4_3x3')
             .conv(1, 1, 256, 1, 1, biased=True, relu=False, name='conv3_4_1x1_increase'))

        (self.feed('conv3_3/relu',
                   'conv3_4_1x1_increase')
             .add(name='conv3_4')
             .relu(name='conv3_4/relu')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_1_1x1_proj'))

        (self.feed('conv3_4/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_1_1x1_reduce')
             .zero_padding(paddings=2, name='padding8')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_1_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_1_1x1_increase'))

        (self.feed('conv4_1_1x1_proj',
                   'conv4_1_1x1_increase')
             .add(name='conv4_1')
             .relu(name='conv4_1/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_2_1x1_reduce')
             .zero_padding(paddings=2, name='padding9')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_2_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_2_1x1_increase'))

        (self.feed('conv4_1/relu',
                   'conv4_2_1x1_increase')
             .add(name='conv4_2')
             .relu(name='conv4_2/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_3_1x1_reduce')
             .zero_padding(paddings=2, name='padding10')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_3_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_3_1x1_increase'))

        (self.feed('conv4_2/relu',
                   'conv4_3_1x1_increase')
             .add(name='conv4_3')
             .relu(name='conv4_3/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_4_1x1_reduce')
             .zero_padding(paddings=2, name='padding11')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_4_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_4_1x1_increase'))

        (self.feed('conv4_3/relu',
                   'conv4_4_1x1_increase')
             .add(name='conv4_4')
             .relu(name='conv4_4/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_5_1x1_reduce')
             .zero_padding(paddings=2, name='padding12')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_5_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_5_1x1_increase'))

        (self.feed('conv4_4/relu',
                   'conv4_5_1x1_increase')
             .add(name='conv4_5')
             .relu(name='conv4_5/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=True, name='conv4_6_1x1_reduce')
             .zero_padding(paddings=2, name='padding13')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=True, name='conv4_6_3x3')
             .conv(1, 1, 512, 1, 1, biased=True, relu=False, name='conv4_6_1x1_increase'))

        (self.feed('conv4_5/relu',
                   'conv4_6_1x1_increase')
             .add(name='conv4_6')
             .relu(name='conv4_6/relu')
             .conv(1, 1, 1024, 1, 1, biased=True, relu=False, name='conv5_1_1x1_proj'))

        (self.feed('conv4_6/relu')
             .conv(1, 1, 256, 1, 1, biased=True, relu=True, name='conv5_1_1x1_reduce')
             .zero_padding(paddings=4, name='padding14')
             .atrous_conv(3, 3, 256, 4, biased=True, relu=True, name='conv5_1_3x3')
             .conv(1, 1, 1024, 1, 1, biased=True, relu=False, name='conv5_1_1x1_increase'))

        (self.feed('conv5_1_1x1_proj',
                   'conv5_1_1x1_increase')
             .add(name='conv5_1')
             .relu(name='conv5_1/relu')
             .conv(1, 1, 256, 1, 1, biased=True, relu=True, name='conv5_2_1x1_reduce')
             .zero_padding(paddings=4, name='padding15')
             .atrous_conv(3, 3, 256, 4, biased=True, relu=True, name='conv5_2_3x3')
             .conv(1, 1, 1024, 1, 1, biased=True, relu=False, name='conv5_2_1x1_increase'))

        (self.feed('conv5_1/relu',
                   'conv5_2_1x1_increase')
             .add(name='conv5_2')
             .relu(name='conv5_2/relu')
             .conv(1, 1, 256, 1, 1, biased=True, relu=True, name='conv5_3_1x1_reduce')
             .zero_padding(paddings=4, name='padding16')
             .atrous_conv(3, 3, 256, 4, biased=True, relu=True, name='conv5_3_3x3')
             .conv(1, 1, 1024, 1, 1, biased=True, relu=False, name='conv5_3_1x1_increase'))

        (self.feed('conv5_2/relu',
                   'conv5_3_1x1_increase')
             .add(name='conv5_3')
             .relu(name='conv5_3/relu'))

        shape = self.layers['conv5_3/relu'].get_shape().as_list()[1:3]
        h, w = shape

        if self.mode == 'eval' and self.cfg.dataset == 'cityscapes': # Change to same configuration as original prototxt
            (self.feed('conv5_3/relu')
                .avg_pool(33, 65, 33, 65, name='conv5_3_pool1')
                .resize_bilinear(shape, name='conv5_3_pool1_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(17, 33, 16, 32, name='conv5_3_pool2')
                .resize_bilinear(shape, name='conv5_3_pool2_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(13, 25, 10, 20, name='conv5_3_pool3')
                .resize_bilinear(shape, name='conv5_3_pool3_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(8, 15, 5, 10, name='conv5_3_pool6')
                .resize_bilinear(shape, name='conv5_3_pool6_interp'))
        else:       # In inference phase, we support different size of images as input.
            (self.feed('conv5_3/relu')
                .avg_pool(h, w, h, w, name='conv5_3_pool1')
                .resize_bilinear(shape, name='conv5_3_pool1_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(h/2, w/2, h/2, w/2, name='conv5_3_pool2')
                .resize_bilinear(shape, name='conv5_3_pool2_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(h/3, w/3, h/3, w/3, name='conv5_3_pool3')
                .resize_bilinear(shape, name='conv5_3_pool3_interp'))

            (self.feed('conv5_3/relu')
                .avg_pool(h/6, w/6, h/6, w/6, name='conv5_3_pool6')
                .resize_bilinear(shape, name='conv5_3_pool6_interp'))

        (self.feed('conv5_3/relu',
                   'conv5_3_pool6_interp',
                   'conv5_3_pool3_interp',
                   'conv5_3_pool2_interp',
                   'conv5_3_pool1_interp')
             .add(name='conv5_3_sum')
             .conv(1, 1, 256, 1, 1, biased=True, relu=True, name='conv5_4_k1')
             .interp(z_factor=2.0, name='conv5_4_interp')
             .zero_padding(paddings=2, name='padding17')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=False, name='conv_sub4'))

        (self.feed('conv3_1/relu')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv3_1_sub2_proj'))

        (self.feed('conv_sub4',
                   'conv3_1_sub2_proj')
             .add(name='sub24_sum')
             .relu(name='sub24_sum/relu')
             .interp(z_factor=2.0, name='sub24_sum_interp')
             .zero_padding(paddings=2, name='padding18')
             .atrous_conv(3, 3, 128, 2, biased=True, relu=False, name='conv_sub2'))

        (self.feed('data')
             .conv(3, 3, 32, 2, 2, biased=True, padding='SAME', relu=True, name='conv1_sub1')
             .conv(3, 3, 32, 2, 2, biased=True, padding='SAME', relu=True, name='conv2_sub1')
             .conv(3, 3, 64, 2, 2, biased=True, padding='SAME', relu=True, name='conv3_sub1')
             .conv(1, 1, 128, 1, 1, biased=True, relu=False, name='conv3_sub1_proj'))

        (self.feed('conv_sub2',
                   'conv3_sub1_proj')
             .add(name='sub12_sum')
             .relu(name='sub12_sum/relu')
             .interp(z_factor=2.0, name='sub12_sum_interp')
             .conv(1, 1, self.cfg.param['num_classes'], 1, 1, biased=True, relu=False, name='conv6_cls'))


class ICNet_BN(Network):
    def __init__(self, cfg, mode, image_reader=None):
        self.cfg = cfg
        self.mode = mode

        if mode == 'train':
            self.images, self.labels = image_reader.next_image, image_reader.next_label    
        
            super().__init__(inputs={'data': self.images}, cfg=self.cfg)

        elif mode == 'eval':
            self.images, self.labels = image_reader.next_image, image_reader.next_label    
        
            super().__init__(inputs={'data': self.images}, cfg=self.cfg)
            
            self.output = self.get_output_node()

        elif mode == 'inference':
            # Create placeholder and pre-process here.
            self.img_placeholder = tf.placeholder(dtype=tf.float32, shape=cfg.INFER_SIZE)
            self.images, self.o_shape, self.n_shape = _infer_preprocess(self.img_placeholder)
            
            super().__init__(inputs={'data': self.images}, cfg=self.cfg)

            self.output = self.get_output_node()

    def get_output_node(self):
        if self.mode == 'inference':
            # Get logits from final layer
            logits = self.layers['conv6_cls']

            # Upscale the logits and decode prediction to get final result.
            logits_up = tf.image.resize_bilinear(logits, size=self.n_shape, align_corners=True)
            logits_up = tf.image.crop_to_bounding_box(logits_up, 0, 0, self.o_shape[0], self.o_shape[1])

            output_classes = tf.argmax(logits_up, axis=3)
            output = decode_labels(output_classes, self.o_shape, self.cfg.param['num_classes'])

        elif self.mode == 'eval':
            logits = self.layers['conv6_cls']

            logits_up = tf.image.resize_bilinear(logits, size=tf.shape(self.labels)[1:3], align_corners=True)
            output = tf.argmax(logits_up, axis=3)
            output = tf.expand_dims(output, axis=3)

        return output

    def predict(self, image):
        return self.sess.run(self.output, feed_dict={self.img_placeholder: image})

    def setup(self):
        (self.feed('data')
             .interp(s_factor=0.5, name='data_sub2')
             .conv(3, 3, 32, 2, 2, biased=False, padding='SAME', relu=False, name='conv1_1_3x3_s2')
             .batch_normalization(relu=True, name='conv1_1_3x3_s2_bn')
             .conv(3, 3, 32, 1, 1, biased=False, padding='SAME', relu=False, name='conv1_2_3x3')
             .batch_normalization(relu=True, name='conv1_2_3x3_bn')
             .conv(3, 3, 64, 1, 1, biased=False, padding='SAME', relu=False, name='conv1_3_3x3')
             .batch_normalization(relu=True, name='conv1_3_3x3_bn')
             .zero_padding(paddings=1, name='padding0')
             .max_pool(3, 3, 2, 2, name='pool1_3x3_s2')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv2_1_1x1_proj')
             .batch_normalization(relu=False, name='conv2_1_1x1_proj_bn'))

        (self.feed('pool1_3x3_s2')
             .conv(1, 1, 32, 1, 1, biased=False, relu=False, name='conv2_1_1x1_reduce')
             .batch_normalization(relu=True, name='conv2_1_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding1')
             .conv(3, 3, 32, 1, 1, biased=False, relu=False, name='conv2_1_3x3')
             .batch_normalization(relu=True, name='conv2_1_3x3_bn')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv2_1_1x1_increase')
             .batch_normalization(relu=False, name='conv2_1_1x1_increase_bn'))

        (self.feed('conv2_1_1x1_proj_bn',
                   'conv2_1_1x1_increase_bn')
             .add(name='conv2_1')
             .relu(name='conv2_1/relu')
             .conv(1, 1, 32, 1, 1, biased=False, relu=False, name='conv2_2_1x1_reduce')
             .batch_normalization(relu=True, name='conv2_2_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding2')
             .conv(3, 3, 32, 1, 1, biased=False, relu=False, name='conv2_2_3x3')
             .batch_normalization(relu=True, name='conv2_2_3x3_bn')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv2_2_1x1_increase')
             .batch_normalization(relu=False, name='conv2_2_1x1_increase_bn'))

        (self.feed('conv2_1/relu',
                   'conv2_2_1x1_increase_bn')
             .add(name='conv2_2')
             .relu(name='conv2_2/relu')
             .conv(1, 1, 32, 1, 1, biased=False, relu=False, name='conv2_3_1x1_reduce')
             .batch_normalization(relu=True, name='conv2_3_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding3')
             .conv(3, 3, 32, 1, 1, biased=False, relu=False, name='conv2_3_3x3')
             .batch_normalization(relu=True, name='conv2_3_3x3_bn')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv2_3_1x1_increase')
             .batch_normalization(relu=False, name='conv2_3_1x1_increase_bn'))

        (self.feed('conv2_2/relu',
                   'conv2_3_1x1_increase_bn')
             .add(name='conv2_3')
             .relu(name='conv2_3/relu')
             .conv(1, 1, 256, 2, 2, biased=False, relu=False, name='conv3_1_1x1_proj')
             .batch_normalization(relu=False, name='conv3_1_1x1_proj_bn'))

        (self.feed('conv2_3/relu')
             .conv(1, 1, 64, 2, 2, biased=False, relu=False, name='conv3_1_1x1_reduce')
             .batch_normalization(relu=True, name='conv3_1_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding4')
             .conv(3, 3, 64, 1, 1, biased=False, relu=False, name='conv3_1_3x3')
             .batch_normalization(relu=True, name='conv3_1_3x3_bn')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv3_1_1x1_increase')
             .batch_normalization(relu=False, name='conv3_1_1x1_increase_bn'))

        (self.feed('conv3_1_1x1_proj_bn',
                   'conv3_1_1x1_increase_bn')
             .add(name='conv3_1')
             .relu(name='conv3_1/relu')
             .interp(s_factor=0.5, name='conv3_1_sub4')
             .conv(1, 1, 64, 1, 1, biased=False, relu=False, name='conv3_2_1x1_reduce')
             .batch_normalization(relu=True, name='conv3_2_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding5')
             .conv(3, 3, 64, 1, 1, biased=False, relu=False, name='conv3_2_3x3')
             .batch_normalization(relu=True, name='conv3_2_3x3_bn')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv3_2_1x1_increase')
             .batch_normalization(relu=False, name='conv3_2_1x1_increase_bn'))

        (self.feed('conv3_1_sub4',
                   'conv3_2_1x1_increase_bn')
             .add(name='conv3_2')
             .relu(name='conv3_2/relu')
             .conv(1, 1, 64, 1, 1, biased=False, relu=False, name='conv3_3_1x1_reduce')
             .batch_normalization(relu=True, name='conv3_3_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding6')
             .conv(3, 3, 64, 1, 1, biased=False, relu=False, name='conv3_3_3x3')
             .batch_normalization(relu=True, name='conv3_3_3x3_bn')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv3_3_1x1_increase')
             .batch_normalization(relu=False, name='conv3_3_1x1_increase_bn'))


        (self.feed('conv3_2/relu',
                   'conv3_3_1x1_increase_bn')
             .add(name='conv3_3')
             .relu(name='conv3_3/relu')
             .conv(1, 1, 64, 1, 1, biased=False, relu=False, name='conv3_4_1x1_reduce')
             .batch_normalization(relu=True, name='conv3_4_1x1_reduce_bn')
             .zero_padding(paddings=1, name='padding7')
             .conv(3, 3, 64, 1, 1, biased=False, relu=False, name='conv3_4_3x3')
             .batch_normalization(relu=True, name='conv3_4_3x3_bn')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv3_4_1x1_increase')
             .batch_normalization(relu=False, name='conv3_4_1x1_increase_bn'))

        (self.feed('conv3_3/relu',
                   'conv3_4_1x1_increase_bn')
             .add(name='conv3_4')
             .relu(name='conv3_4/relu')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_1_1x1_proj')
             .batch_normalization(relu=False, name='conv4_1_1x1_proj_bn'))

        (self.feed('conv3_4/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_1_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_1_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding8')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_1_3x3')
             .batch_normalization(relu=True, name='conv4_1_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_1_1x1_increase')
             .batch_normalization(relu=False, name='conv4_1_1x1_increase_bn'))

        (self.feed('conv4_1_1x1_proj_bn',
                   'conv4_1_1x1_increase_bn')
             .add(name='conv4_1')
             .relu(name='conv4_1/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_2_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_2_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding9')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_2_3x3')
             .batch_normalization(relu=True, name='conv4_2_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_2_1x1_increase')
             .batch_normalization(relu=False, name='conv4_2_1x1_increase_bn'))

        (self.feed('conv4_1/relu',
                   'conv4_2_1x1_increase_bn')
             .add(name='conv4_2')
             .relu(name='conv4_2/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_3_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_3_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding10')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_3_3x3')
             .batch_normalization(relu=True, name='conv4_3_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_3_1x1_increase')
             .batch_normalization(relu=False, name='conv4_3_1x1_increase_bn'))

        (self.feed('conv4_2/relu',
                   'conv4_3_1x1_increase_bn')
             .add(name='conv4_3')
             .relu(name='conv4_3/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_4_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_4_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding11')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_4_3x3')
             .batch_normalization(relu=True, name='conv4_4_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_4_1x1_increase')
             .batch_normalization(relu=False, name='conv4_4_1x1_increase_bn'))

        (self.feed('conv4_3/relu',
                   'conv4_4_1x1_increase_bn')
             .add(name='conv4_4')
             .relu(name='conv4_4/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_5_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_5_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding12')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_5_3x3')
             .batch_normalization(relu=True, name='conv4_5_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_5_1x1_increase')
             .batch_normalization(relu=False, name='conv4_5_1x1_increase_bn'))

        (self.feed('conv4_4/relu',
                   'conv4_5_1x1_increase_bn')
             .add(name='conv4_5')
             .relu(name='conv4_5/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv4_6_1x1_reduce')
             .batch_normalization(relu=True, name='conv4_6_1x1_reduce_bn')
             .zero_padding(paddings=2, name='padding13')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv4_6_3x3')
             .batch_normalization(relu=True, name='conv4_6_3x3_bn')
             .conv(1, 1, 512, 1, 1, biased=False, relu=False, name='conv4_6_1x1_increase')
             .batch_normalization(relu=False, name='conv4_6_1x1_increase_bn'))

        (self.feed('conv4_5/relu',
                   'conv4_6_1x1_increase_bn')
             .add(name='conv4_6')
             .relu(name='conv4_6/relu')
             .conv(1, 1, 1024, 1, 1, biased=False, relu=False, name='conv5_1_1x1_proj')
             .batch_normalization(relu=False, name='conv5_1_1x1_proj_bn'))

        (self.feed('conv4_6/relu')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv5_1_1x1_reduce')
             .batch_normalization(relu=True, name='conv5_1_1x1_reduce_bn')
             .zero_padding(paddings=4, name='padding14')
             .atrous_conv(3, 3, 256, 4, biased=False, relu=False, name='conv5_1_3x3')
             .batch_normalization(relu=True, name='conv5_1_3x3_bn')
             .conv(1, 1, 1024, 1, 1, biased=False, relu=False, name='conv5_1_1x1_increase')
             .batch_normalization(relu=False, name='conv5_1_1x1_increase_bn'))

        (self.feed('conv5_1_1x1_proj_bn',
                   'conv5_1_1x1_increase_bn')
             .add(name='conv5_1')
             .relu(name='conv5_1/relu')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv5_2_1x1_reduce')
             .batch_normalization(relu=True, name='conv5_2_1x1_reduce_bn')
             .zero_padding(paddings=4, name='padding15')
             .atrous_conv(3, 3, 256, 4, biased=False, relu=False, name='conv5_2_3x3')
             .batch_normalization(relu=True, name='conv5_2_3x3_bn')
             .conv(1, 1, 1024, 1, 1, biased=False, relu=False, name='conv5_2_1x1_increase')
             .batch_normalization(relu=False, name='conv5_2_1x1_increase_bn'))

        (self.feed('conv5_1/relu',
                   'conv5_2_1x1_increase_bn')
             .add(name='conv5_2')
             .relu(name='conv5_2/relu')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv5_3_1x1_reduce')
             .batch_normalization(relu=True, name='conv5_3_1x1_reduce_bn')
             .zero_padding(paddings=4, name='padding16')
             .atrous_conv(3, 3, 256, 4, biased=False, relu=False, name='conv5_3_3x3')
             .batch_normalization(relu=True, name='conv5_3_3x3_bn')
             .conv(1, 1, 1024, 1, 1, biased=False, relu=False, name='conv5_3_1x1_increase')
             .batch_normalization(relu=False, name='conv5_3_1x1_increase_bn'))

        (self.feed('conv5_2/relu',
                   'conv5_3_1x1_increase_bn')
             .add(name='conv5_3')
             .relu(name='conv5_3/relu'))

        shape = self.layers['conv5_3/relu'].get_shape().as_list()[1:3]
        h, w = shape

        (self.feed('conv5_3/relu')
             .avg_pool(h, w, h, w, name='conv5_3_pool1')
             .resize_bilinear(shape, name='conv5_3_pool1_interp'))

        (self.feed('conv5_3/relu')
             .avg_pool(h/2, w/2, h/2, w/2, name='conv5_3_pool2')
             .resize_bilinear(shape, name='conv5_3_pool2_interp'))

        (self.feed('conv5_3/relu')
             .avg_pool(h/3, w/3, h/3, w/3, name='conv5_3_pool3')
             .resize_bilinear(shape, name='conv5_3_pool3_interp'))

        (self.feed('conv5_3/relu')
             .avg_pool(h/4, w/4, h/4, w/4, name='conv5_3_pool6')
             .resize_bilinear(shape, name='conv5_3_pool6_interp'))

        (self.feed('conv5_3/relu',
                   'conv5_3_pool6_interp',
                   'conv5_3_pool3_interp',
                   'conv5_3_pool2_interp',
                   'conv5_3_pool1_interp')
             .add(name='conv5_3_sum')
             .conv(1, 1, 256, 1, 1, biased=False, relu=False, name='conv5_4_k1')
             .batch_normalization(relu=True, name='conv5_4_k1_bn')
             .interp(z_factor=2.0, name='conv5_4_interp')
             .zero_padding(paddings=2, name='padding17')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv_sub4')
             .batch_normalization(relu=False, name='conv_sub4_bn'))

        (self.feed('conv3_1/relu')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv3_1_sub2_proj')
             .batch_normalization(relu=False, name='conv3_1_sub2_proj_bn'))

        (self.feed('conv_sub4_bn',
                   'conv3_1_sub2_proj_bn')
             .add(name='sub24_sum')
             .relu(name='sub24_sum/relu')
             .interp(z_factor=2.0, name='sub24_sum_interp')
             .zero_padding(paddings=2, name='padding18')
             .atrous_conv(3, 3, 128, 2, biased=False, relu=False, name='conv_sub2')
             .batch_normalization(relu=False, name='conv_sub2_bn'))

        (self.feed('data')
             .conv(3, 3, 32, 2, 2, biased=False, padding='SAME', relu=False, name='conv1_sub1')
             .batch_normalization(relu=True, name='conv1_sub1_bn')
             .conv(3, 3, 32, 2, 2, biased=False, padding='SAME', relu=False, name='conv2_sub1')
             .batch_normalization(relu=True, name='conv2_sub1_bn')
             .conv(3, 3, 64, 2, 2, biased=False, padding='SAME', relu=False, name='conv3_sub1')
             .batch_normalization(relu=True, name='conv3_sub1_bn')
             .conv(1, 1, 128, 1, 1, biased=False, relu=False, name='conv3_sub1_proj')
             .batch_normalization(relu=False, name='conv3_sub1_proj_bn'))

        (self.feed('conv_sub2_bn',
                   'conv3_sub1_proj_bn')
             .add(name='sub12_sum')
             .relu(name='sub12_sum/relu')
             .interp(z_factor=2.0, name='sub12_sum_interp')
             .conv(1, 1, self.cfg.param['num_classes'], 1, 1, biased=True, relu=False, name='conv6_cls'))

        (self.feed('conv5_4_interp')
             .conv(1, 1, self.cfg.param['num_classes'], 1, 1, biased=True, relu=False, name='sub4_out'))

        (self.feed('sub24_sum_interp')
             .conv(1, 1, self.cfg.param['num_classes'], 1, 1, biased=True, relu=False, name='sub24_out'))