Merge branch 'main' into unsupervised

packnet_functional.py

@@ -22,10 +22,10 @@ def residual_layer(inputs, out_channels, stride, dropout=None):
     x = layers.Conv2D(out_channels, 3, padding='same', strides=stride)(inputs)
     x = layers.Conv2D(out_channels, 3, padding='same')(x)
     shortcut = layers.Conv2D(
-        out_channels, 3, padding='same', strides=stride)(inputs)
+        out_channels, 1, padding='same', strides=stride)(inputs)
     if dropout:
         shortcut = keras.layers.SpatialDropout2D(dropout)(shortcut)
-    x = keras.layers.Concatenate()([x, shortcut])
+    x = keras.layers.Add()([x, shortcut])
     x = group_norm.GroupNormalization(16)(x)
     return keras.layers.ELU()(x)
 
@@ -52,11 +52,11 @@ def packnet_inverse_depth(inputs, out_channels=1, min_depth=0.5):
 
 def pack_3d(inputs, kernel_size, r=2, features_3d=8):
     """
-    Implementatino of the 3d packing block proposed here: https://arxiv.org/abs/1905.02693
-    :param inputs:
-    :param kernel_size:
-    :param r:
-    :param features_3d:
+    Implementation of the 3d packing block proposed here: https://arxiv.org/abs/1905.02693
+    :param inputs: Tensor inputs
+    :param kernel_size: Conv3D kernels size
+    :param r: Packing factor
+    :param features_3d: Packing depth (increase to increase number of parameters and accuracy)
     :return:
     """
     # Data format for single image in nyu is HWC (space_to_depth uses NHWC as default)
@@ -78,19 +78,23 @@ def unpack_3d(inputs, out_channels, kernel_size, r=2, features_3d=8):
     return nn.depth_to_space(x, r)
 
 
-# TODO: Support different size packnet for scaling up/down
 # TODO: Support different channel format (right now we're supporting NHWC, we should also support NCHW)
-def make_packnet(shape=(224, 224, 3), skip_add=True, features_3d=4, dropout=None):
+def make_packnet(shape=(224, 224, 3), skip_add=False, features_3d=8, dropout=None, small=False):
     """
     Make the PackNet depth network.
     :param shape: Input shape of the image
     :param skip_add: Set to use add rather than concat skip connections, defaults to True
-    :return:
+    :param features_3d: Number of layers in 3D conv for packing/unpacking layers
+    :param dropout: Whether to build the model with dropout layers fpr regularisation. Useful in training only
+    :param small: Set True to not include the middle-most layer. Reduces params from ~128M -> ~34M
+    Further reductions can be achieved by using additive skip connections and less 3d features (down to min ~10M)
+    :return: Packnet Keras Model
     """
 
     # ================ ENCODER =================
     input = keras.layers.Input(shape=shape)
-    x = packnet_conv2d(input, 32, 5, 1)
+    initial_conv_channels = 32 if small else 64
+    x = packnet_conv2d(input, initial_conv_channels, 5, 1)
     skip_1 = x
     x = packnet_conv2d(x, 64, 7, 1)
     x = pack_3d(x, 5, features_3d=features_3d)
@@ -104,18 +108,22 @@ def make_packnet(shape=(224, 224, 3), skip_add=True, features_3d=4, dropout=None
     x = residual_block(x, 256, 3, 1, dropout)
     x = pack_3d(x, 3, features_3d=features_3d)
     skip_5 = x
+    if not small:
         x = residual_block(x, 512, 3, 1, dropout)
         x = pack_3d(x, 3, features_3d=features_3d)
     # ================ ENCODER =================
 
     # ================ DECODER =================
+    # Addition requires we half the outputs so there is a matching number of channels
+    divide_factor = (2 if skip_add else 1)
     # layer 7
-    x = unpack_3d(x, 512, 3, features_3d=features_3d)
+    if not small:
+        x = unpack_3d(x, 512 // divide_factor, 3, features_3d=features_3d)
         x = keras.layers.Add()(
             [x, skip_5]) if skip_add else keras.layers.Concatenate()([x, skip_5])
         x = packnet_conv2d(x, 512, 3, 1)
     # layer 8
-    x = unpack_3d(x, 256, 3, features_3d=features_3d)
+    x = unpack_3d(x, 256 // divide_factor, 3, features_3d=features_3d)
     x = keras.layers.Add()(
         [x, skip_4]) if skip_add else keras.layers.Concatenate()([x, skip_4])
     x = packnet_conv2d(x, 256, 3, 1)
@@ -123,7 +131,7 @@ def make_packnet(shape=(224, 224, 3), skip_add=True, features_3d=4, dropout=None
     # layer 9
     x = packnet_inverse_depth(x, 1)
     # layer 10
-    u_layer_8 = unpack_3d(layer_8, 128, 3, features_3d=features_3d)
+    u_layer_8 = unpack_3d(layer_8, 128 // divide_factor, 3, features_3d=features_3d)
     x = keras.layers.UpSampling2D()(x)
     x = keras.layers.Add()([u_layer_8, skip_3, x]) if skip_add else keras.layers.Concatenate()([u_layer_8, skip_3, x])
     x = packnet_conv2d(x, 128, 3, 1)
@@ -139,12 +147,20 @@ def make_packnet(shape=(224, 224, 3), skip_add=True, features_3d=4, dropout=None
     # layer 13
     x = packnet_inverse_depth(x)
     # layer 14
-    u_layer_12 = unpack_3d(layer_12, 32, 3, features_3d=features_3d)
+    u_layer_12 = unpack_3d(layer_12, initial_conv_channels, 3, features_3d=features_3d)
     x = keras.layers.UpSampling2D()(x)
     x = keras.layers.Add()([u_layer_12, skip_1, x]) if skip_add else keras.layers.Concatenate()([u_layer_12, skip_1, x])
-    x = packnet_conv2d(x, 32, 3, 1)
+    x = packnet_conv2d(x, initial_conv_channels, 3, 1)
     # layer 15
     x = packnet_inverse_depth(x)
     # ================ DECODER =================
 
     return keras.Model(inputs=input, outputs=x, name="PackNet")
+
+
+if __name__ == '__main__':
+    # This is the implementation used by the packnet sfm paper
+    make_packnet().summary()
+
+    # This is the very small version of packnet
+    make_packnet(small=True, features_3d=1, skip_add=True).summary()