chore: Include license

keras_layer_AnchorBoxes.py

@@ -1,3 +1,22 @@
+'''
+A custom Keras layer to generate anchor boxes.
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+'''
+
 import keras.backend as K
 from keras.engine.topology import InputSpec
 from keras.engine.topology import Layer
@@ -46,6 +65,7 @@ def __init__(self,
                  limit_boxes=True,
                  variances=[1.0, 1.0, 1.0, 1.0],
                  coords='centroids',
+                 normalize_coords=False,
                  **kwargs):
         '''
         All arguments need to be set to the same values used to train the model, otherwise the behavior is undefined.
@@ -73,6 +93,8 @@ def __init__(self,
             coords (str, optional): The box coordinate format to be used. Can be either 'centroids' for the format
                 `(cx, cy, w, h)` (box center coordinates, width, and height) or 'minmax' for the format
                 `(xmin, xmax, ymin, ymax)`. Defaults to 'centroids'.
+            normalize_coords (bool, optional): Set to `True` if the model uses relative instead of absolute coordinates,
+                i.e. if the model predicts box coordinates within [0,1] instead of absolute coordinates. Defaults to `False`.
         '''
         if K.backend() != 'tensorflow':
             raise TypeError("This layer only supports TensorFlow at the moment, but you are using the {} backend.".format(K.backend()))
@@ -95,6 +117,7 @@ def __init__(self,
         self.limit_boxes = limit_boxes
         self.variances = variances
         self.coords = coords
+        self.normalize_coords = normalize_coords
         # Compute the number of boxes per cell
         if (1 in aspect_ratios) & two_boxes_for_ar1:
             self.n_boxes = len(aspect_ratios) + 1
@@ -179,6 +202,11 @@ def call(self, x, mask=None):
             y_coords[y_coords < 0] = 0
             boxes_tensor[:,:,:,[2, 3]] = y_coords
 
+        # `normalize_coords` is enabled, normalize the coordinates to be within [0,1]
+        if self.normalize_coords:
+            boxes_tensor[:, :, :, :2] /= self.img_width
+            boxes_tensor[:, :, :, 2:] /= self.img_height
+
         if self.coords == 'centroids':
             # TODO: Implement box limiting directly for `(cx, cy, w, h)` so that we don't have to unnecessarily convert back and forth
             # Convert `(xmin, xmax, ymin, ymax)` back to `(cx, cy, w, h)`

keras_layer_L2Normalization.py

@@ -1,3 +1,22 @@
+'''
+A custom Keras layer to perform L2-normalization.
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+'''
+
 import keras.backend as K
 from keras.engine.topology import InputSpec
 from keras.engine.topology import Layer

keras_ssd300.py

@@ -1,3 +1,22 @@
+'''
+A Keras port of the original Caffe SSD300 network.
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+'''
+
 import numpy as np
 from keras.models import Model
 from keras.layers import Input, Lambda, Activation, Conv2D, MaxPooling2D, Reshape, Concatenate
@@ -20,7 +39,8 @@ def ssd_300(image_size,
             two_boxes_for_ar1=True,
             limit_boxes=False,
             variances=[0.1, 0.1, 0.2, 0.2],
-            coords='centroids'):
+            coords='centroids',
+            normalize_coords=False):
     '''
     Build a Keras model with SSD_300 architecture, see references.
 
@@ -72,11 +92,13 @@ def ssd_300(image_size,
         variances (list, optional): A list of 4 floats >0 with scaling factors (actually it's not factors but divisors
             to be precise) for the encoded predicted box coordinates. A variance value of 1.0 would apply
             no scaling at all to the predictions, while values in (0,1) upscale the encoded predictions and values greater
-            than 1.0 downscale the encoded predictions. Defaults to `[0.1, 0.1, 0.2, 0.2]`, following the original
-            implementation. The coordinate format must be 'centroids'.
+            than 1.0 downscale the encoded predictions. Defaults to `[0.1, 0.1, 0.2, 0.2]`, following the original implementation.
+            The coordinate format must be 'centroids'.
         coords (str, optional): The box coordinate format to be used. Can be either 'centroids' for the format
             `(cx, cy, w, h)` (box center coordinates, width, and height) or 'minmax' for the format
             `(xmin, xmax, ymin, ymax)`. Defaults to 'centroids', following the original implementation.
+        normalize_coords (bool, optional): Set to `True` if the model is supposed to use relative instead of absolute coordinates,
+            i.e. if the model predicts box coordinates within [0,1] instead of absolute coordinates. Defaults to `False`.
 
     Returns:
         model: The Keras SSD model.
@@ -162,7 +184,7 @@ def ssd_300(image_size,
     ### Design the actual network
 
     x = Input(shape=(img_height, img_width, img_channels))
-    normed = Lambda(lambda z: z/127.5 - 1., # Convert input feature range to [-1,1]
+    normed = Lambda(lambda z: z/127.5 - 1.0, # Convert input feature range to [-1,1]
                     output_shape=(img_height, img_width, img_channels),
                     name='lambda1')(x)
 
@@ -231,17 +253,17 @@ def ssd_300(image_size,
 
     # Output shape of anchors: `(batch, height, width, n_boxes, 8)`
     conv4_3_norm_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[0], next_scale=scales[1], aspect_ratios=aspect_ratios_conv4_3,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='conv4_3_norm_mbox_priorbox')(conv4_3_norm)
+                                             two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='conv4_3_norm_mbox_priorbox')(conv4_3_norm)
     fc7_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[1], next_scale=scales[2], aspect_ratios=aspect_ratios_fc7,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='fc7_mbox_priorbox')(fc7)
+                                    two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='fc7_mbox_priorbox')(fc7)
     conv6_2_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[2], next_scale=scales[3], aspect_ratios=aspect_ratios_conv6_2,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='conv6_2_mbox_priorbox')(conv6_2)
+                                        two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='conv6_2_mbox_priorbox')(conv6_2)
     conv7_2_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[3], next_scale=scales[4], aspect_ratios=aspect_ratios_conv7_2,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='conv7_2_mbox_priorbox')(conv7_2)
+                                        two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='conv7_2_mbox_priorbox')(conv7_2)
     conv8_2_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[4], next_scale=scales[5], aspect_ratios=aspect_ratios_conv8_2,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='conv8_2_mbox_priorbox')(conv8_2)
+                                        two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='conv8_2_mbox_priorbox')(conv8_2)
     conv9_2_mbox_priorbox = AnchorBoxes(img_height, img_width, this_scale=scales[5], next_scale=scales[6], aspect_ratios=aspect_ratios_conv9_2,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='conv9_2_mbox_priorbox')(conv9_2)
+                                        two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='conv9_2_mbox_priorbox')(conv9_2)
 
     ### Reshape
 

keras_ssd7.py

@@ -1,3 +1,22 @@
+'''
+A small Keras model with SSD architecture.
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+'''
+
 import numpy as np
 from keras.models import Model
 from keras.layers import Input, Lambda, Convolution2D, MaxPooling2D, BatchNormalization, ELU, Reshape, Concatenate, Activation
@@ -14,7 +33,8 @@ def build_model(image_size,
                 two_boxes_for_ar1=True,
                 limit_boxes=True,
                 variances=[1.0, 1.0, 1.0, 1.0],
-                coords='centroids'):
+                coords='centroids',
+                normalize_coords=False):
     '''
     Build a Keras model with SSD architecture, see references.
 
@@ -72,6 +92,8 @@ def build_model(image_size,
         coords (str, optional): The box coordinate format to be used. Can be either 'centroids' for the format
             `(cx, cy, w, h)` (box center coordinates, width, and height) or 'minmax' for the format
             `(xmin, xmax, ymin, ymax)`. Defaults to 'centroids'.
+        normalize_coords (bool, optional): Set to `True` if the model is supposed to use relative instead of absolute coordinates,
+            i.e. if the model predicts box coordinates within [0,1] instead of absolute coordinates. Defaults to `False`.
 
     Returns:
         model: The Keras SSD model.
@@ -152,7 +174,6 @@ def build_model(image_size,
                     output_shape=(img_height, img_width, img_channels),
                     name='lambda1')(x)
 
-
     conv1 = Convolution2D(32, (5, 5), name='conv1', strides=(1, 1), padding="same")(normed)
     conv1 = BatchNormalization(axis=3, momentum=0.99, name='bn1')(conv1) # Tensorflow uses filter format [filter_height, filter_width, in_channels, out_channels], hence axis = 3
     conv1 = ELU(name='elu1')(conv1)
@@ -204,13 +225,13 @@ def build_model(image_size,
     # Generate the anchor boxes
     # Output shape of anchors: `(batch, height, width, n_boxes, 8)`
     anchors4 = AnchorBoxes(img_height, img_width, this_scale=scales[0], next_scale=scales[1], aspect_ratios=aspect_ratios_conv4,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='anchors4')(boxes4)
+                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='anchors4')(boxes4)
     anchors5 = AnchorBoxes(img_height, img_width, this_scale=scales[1], next_scale=scales[2], aspect_ratios=aspect_ratios_conv5,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='anchors5')(boxes5)
+                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='anchors5')(boxes5)
     anchors6 = AnchorBoxes(img_height, img_width, this_scale=scales[2], next_scale=scales[3], aspect_ratios=aspect_ratios_conv6,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='anchors6')(boxes6)
+                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='anchors6')(boxes6)
     anchors7 = AnchorBoxes(img_height, img_width, this_scale=scales[3], next_scale=scales[4], aspect_ratios=aspect_ratios_conv7,
-                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, name='anchors7')(boxes7)
+                           two_boxes_for_ar1=two_boxes_for_ar1, limit_boxes=limit_boxes, variances=variances, coords=coords, normalize_coords=normalize_coords, name='anchors7')(boxes7)
 
     # Reshape the class predictions, yielding 3D tensors of shape `(batch, height * width * n_boxes, n_classes)`
     # We want the classes isolated in the last axis to perform softmax on them

keras_ssd_loss.py

@@ -1,4 +1,21 @@
-'''The Keras-compatible loss function for the SSD model. Currently supports TensorFlow only.'''
+'''
+The Keras-compatible loss function for the SSD model. Currently supports TensorFlow only.
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
+'''
 
 import tensorflow as tf
 

ssd_batch_generator.py

@@ -2,6 +2,21 @@
 Includes:
 * A batch generator for SSD model training and inference which can perform online data agumentation
 * An offline image processor that saves processed images and adjusted labels to disk
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
 '''
 
 import numpy as np

ssd_box_encode_decode_utils.py

@@ -1,8 +1,24 @@
 '''
 Includes:
-* Function to compute IoU similarity
-* Function to decode SSD model output
+* Function to compute IoU similarity for axis-aligned, rectangular, 2D bounding boxes
+* Function to perform greedy non-maximum suppression
+* Function to decode raw SSD model output
 * Class to encode targets for SSD model training
+
+Copyright (C) 2017 Pierluigi Ferrari
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program.  If not, see <http://www.gnu.org/licenses/>.
 '''
 
 import numpy as np