Update model.py

mrcnn/model.py

@@ -31,6 +31,18 @@
 assert LooseVersion(keras.__version__) >= LooseVersion('2.0.8')
 
 
+class AnchorsLayer(KL.Layer):
+    def __init__(self, anchors, name="anchors", **kwargs):
+        super(AnchorsLayer, self).__init__(name=name, **kwargs)
+        self.anchors = tf.Variable(anchors)
+
+    def call(self, dummy):
+        return self.anchors
+
+    def get_config(self):
+        config = super(AnchorsLayer, self).get_config()
+        return config
+
 ############################################################
 #  Utility Functions
 ############################################################
@@ -338,7 +350,7 @@ def compute_output_shape(self, input_shape):
 
 def log2_graph(x):
     """Implementation of Log2. TF doesn't have a native implementation."""
-    return tf.log(x) / tf.log(2.0)
+    return tf.math.log(x) / tf.math.log(2.0)
 
 
 class PyramidROIAlign(KE.Layer):
@@ -550,12 +562,12 @@ def detection_targets_graph(proposals, gt_class_ids, gt_boxes, gt_masks, config)
     # Positive ROIs
     positive_count = int(config.TRAIN_ROIS_PER_IMAGE *
                          config.ROI_POSITIVE_RATIO)
-    positive_indices = tf.random_shuffle(positive_indices)[:positive_count]
+    positive_indices = tf.random.shuffle(positive_indices)[:positive_count]
     positive_count = tf.shape(positive_indices)[0]
     # Negative ROIs. Add enough to maintain positive:negative ratio.
     r = 1.0 / config.ROI_POSITIVE_RATIO
     negative_count = tf.cast(r * tf.cast(positive_count, tf.float32), tf.int32) - positive_count
-    negative_indices = tf.random_shuffle(negative_indices)[:negative_count]
+    negative_indices = tf.random.shuffle(negative_indices)[:negative_count]
     # Gather selected ROIs
     positive_rois = tf.gather(proposals, positive_indices)
     negative_rois = tf.gather(proposals, negative_indices)
@@ -1931,7 +1943,7 @@ def build(self, mode, config):
             # TODO: can this be optimized to avoid duplicating the anchors?
             anchors = np.broadcast_to(anchors, (config.BATCH_SIZE,) + anchors.shape)
             # A hack to get around Keras's bad support for constants
-            anchors = KL.Lambda(lambda x: tf.Variable(anchors), name="anchors")(input_image)
+            anchors = AnchorsLayer(anchors, name="anchors")(input_image)
         else:
             anchors = input_anchors
 
@@ -2112,7 +2124,7 @@ def load_weights(self, filepath, by_name=False, exclude=None):
 
         if h5py is None:
             raise ImportError('`load_weights` requires h5py.')
-        f = h5py.File(filepath, mode='r')
+        f = h5py.save(filepath, mode='r')
         if 'layer_names' not in f.attrs and 'model_weights' in f:
             f = f['model_weights']
 

mrcnn/utils.py

@@ -199,8 +199,8 @@ def box_refinement_graph(box, gt_box):
 
     dy = (gt_center_y - center_y) / height
     dx = (gt_center_x - center_x) / width
-    dh = tf.log(gt_height / height)
-    dw = tf.log(gt_width / width)
+    dh = tf.math.log(gt_height / height)
+    dw = tf.math.log(gt_width / width)
 
     result = tf.stack([dy, dx, dh, dw], axis=1)
     return result