Added the DenseASPP model from CVPR 2018

builders/model_builder.py

@@ -14,9 +14,11 @@
 from models.DeepLabV3_plus import build_deeplabv3_plus
 from models.AdapNet import build_adaptnet
 from models.custom_model import build_custom
+from models.DenseASPP import build_dense_aspp
 
 SUPPORTED_MODELS = ["FC-DenseNet56", "FC-DenseNet67", "FC-DenseNet103", "Encoder-Decoder", "Encoder-Decoder-Skip", "RefineNet",
-    "FRRN-A", "FRRN-B", "MobileUNet", "MobileUNet-Skip", "PSPNet", "GCN", "DeepLabV3", "DeepLabV3_plus", "AdapNet", "custom"]
+    "FRRN-A", "FRRN-B", "MobileUNet", "MobileUNet-Skip", "PSPNet", "GCN", "DeepLabV3", "DeepLabV3_plus", "AdapNet", 
+    "DenseASPP", "custom"]
 
 SUPPORTED_FRONTENDS = ["ResNet50", "ResNet101", "ResNet152", "MobileNetV2", "InceptionV4"]
 
@@ -32,10 +34,10 @@ def build_model(model_name, net_input, num_classes, crop_width, crop_height, fro
 	print("Preparing the model ...")
 
 	if model_name not in SUPPORTED_MODELS:
-		raise ValueError("The model you selelect is not supported. The following models are currently supported: {0}".format(SUPPORTED_MODELS))
+		raise ValueError("The model you selected is not supported. The following models are currently supported: {0}".format(SUPPORTED_MODELS))
 
 	if frontend not in SUPPORTED_FRONTENDS:
-		raise ValueError("The frontend you selelect is not supported. The following models are currently supported: {0}".format(SUPPORTED_FRONTENDS))
+		raise ValueError("The frontend you selected is not supported. The following models are currently supported: {0}".format(SUPPORTED_FRONTENDS))
 
 	if "ResNet50" == frontend and not os.path.isfile("models/resnet_v2_50.ckpt"):
 	    download_checkpoints("ResNet50")
@@ -74,6 +76,9 @@ def build_model(model_name, net_input, num_classes, crop_width, crop_height, fro
 	elif model_name == "DeepLabV3_plus":
 	    # DeepLabV3+ requires pre-trained ResNet weights
 	    network, init_fn = build_deeplabv3_plus(net_input, preset_model = model_name, frontend=frontend, num_classes=num_classes, is_training=is_training)
+	elif model_name == "DenseASPP":
+	    # DenseASPP+ requires pre-trained ResNet weights
+	    network, init_fn = build_dense_aspp(net_input, preset_model = model_name, frontend=frontend, num_classes=num_classes, is_training=is_training)
 	elif model_name == "AdapNet":
 	    network = build_adaptnet(net_input, num_classes=num_classes)
 	elif model_name == "custom":

docs/README.md

@@ -4,21 +4,21 @@
 
 ## News
 
-**What's New:**
+### What's New
 
-- Plotting for every epoch, similar to Tensorboard
+- Added the DenseASPP network from CVPR 2018!
 
 - Added support for MobileNetV2 and InceptionV4 frontends!
 
 - Code restructuring. Much easier to expand and debug **You can now set the segmentation model and frontend to use (ResNet50, ResNet101, etc) separately as command line arguments. See the updated usage section below**
 
-**Coming Soon:**
+### Coming Soon
 
-- Anything that comes out at CVPR 2018!
+- Anything that comes out at CVPR 2018 and ECCV 2018!
 
-- Support for exporting inference graph.
+- More network frontends!
 
-Open up an issue to suggest a new feature or improvement!
+**Open up an issue to suggest a new feature or improvement!**
 
 ## Description
 This repository serves as a Semantic Segmentation Suite. The goal is to easily be able to implement, train, and test new Semantic Segmentation models! Complete with the following:
@@ -73,6 +73,8 @@ to obtain robust features for recognition. The two streams are coupled at the fu
 
 - [Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation](https://arxiv.org/abs/1802.02611). This is the DeepLabV3+ network which adds a Decoder module on top of the regular DeepLabV3 model.
 
+- [DenseASPP for Semantic Segmentation in Street Scenes](http://openaccess.thecvf.com/content_cvpr_2018/html/Yang_DenseASPP_for_Semantic_CVPR_2018_paper.html). Combines many different scales using dilated convolution but with dense connections
+
 - Or make your own and plug and play!
 
 **Note:** If you are using any of the networks that rely on a pre-trained ResNet, then you will need to download the pre-trained weights using the provided script. These are currently: PSPNet, RefineNet, DeepLabV3, DeepLabV3+, GCN.

models/DenseASPP.py

@@ -0,0 +1,60 @@
+import tensorflow as tf
+from tensorflow.contrib import slim
+from builders import frontend_builder
+import os, sys
+
+
+def Upsampling(inputs,scale):
+    return tf.image.resize_nearest_neighbor(inputs, size=[tf.shape(inputs)[1]*scale,  tf.shape(inputs)[2]*scale])
+
+
+
+def DilatedConvBlock(inputs, n_filters, rate=1, kernel_size=[3, 3]):
+    """
+    Basic dilated conv block 
+    Apply successivly BatchNormalization, ReLU nonlinearity, dilated convolution 
+    """
+    net = tf.nn.relu(slim.batch_norm(inputs, fused=True))
+    net = slim.conv2d(net, n_filters, kernel_size, rate=rate, activation_fn=None, normalizer_fn=None)
+    return net
+
+
+
+def build_dense_aspp(inputs, num_classes, preset_model='DenseASPP', frontend="ResNet101", weight_decay=1e-5, is_training=True, pretrained_dir="models"):
+
+
+    logits, end_points, frontend_scope, init_fn  = frontend_builder.build_frontend(inputs, frontend, is_training=is_training)
+
+    init_features = end_points['pool3']
+
+    ### First block, rate = 3
+    d_3_features = DilatedConvBlock(init_features, n_filters=256, kernel_size=[1, 1])
+    d_3 = DilatedConvBlock(d_3_features, n_filters=64, rate=3, kernel_size=[3, 3])
+
+    ### Second block, rate = 6
+    d_4 = tf.concat([init_features, d_3], axis=-1)
+    d_4 = DilatedConvBlock(d_4, n_filters=256, kernel_size=[1, 1])
+    d_4 = DilatedConvBlock(d_4, n_filters=64, rate=6, kernel_size=[3, 3])
+
+    ### Third block, rate = 12
+    d_5 = tf.concat([init_features, d_3, d_4], axis=-1)
+    d_5 = DilatedConvBlock(d_5, n_filters=256, kernel_size=[1, 1])
+    d_5 = DilatedConvBlock(d_5, n_filters=64, rate=12, kernel_size=[3, 3])
+
+    ### Fourth block, rate = 18
+    d_6 = tf.concat([init_features, d_3, d_4, d_5], axis=-1)
+    d_6 = DilatedConvBlock(d_6, n_filters=256, kernel_size=[1, 1])
+    d_6 = DilatedConvBlock(d_6, n_filters=64, rate=18, kernel_size=[3, 3])
+
+    ### Fifth block, rate = 24
+    d_7 = tf.concat([init_features, d_3, d_4, d_5, d_6], axis=-1)
+    d_7 = DilatedConvBlock(d_7, n_filters=256, kernel_size=[1, 1])
+    d_7 = DilatedConvBlock(d_7, n_filters=64, rate=24, kernel_size=[3, 3])
+
+    full_block = tf.concat([init_features, d_3, d_4, d_5, d_6, d_7], axis=-1)
+
+    net = slim.conv2d(full_block, num_classes, [1, 1], activation_fn=None, scope='logits')
+
+    net = Upsampling(net, scale=8)
+
+    return net, init_fn