pr

research/object_detection/core/box_list_ops_test.py

@@ -217,7 +217,6 @@ class BoxListOpsTest(test_case.TestCase):
 
   def test_iou(self):
     def graph_fn():
-      corners1 = tf.constant([[4.0, 3.0, 7.0, 5.0], [5.0, 6.0, 10.0, 7.0]])
       corners1 = tf.constant([[4.0, 3.0, 7.0, 5.0], [5.0, 6.0, 10.0, 7.0]])
       corners2 = tf.constant([[3.0, 4.0, 6.0, 8.0], [14.0, 14.0, 15.0, 15.0],
                               [0.0, 0.0, 20.0, 20.0]])
@@ -229,6 +228,31 @@ class BoxListOpsTest(test_case.TestCase):
     iou_output = self.execute(graph_fn, [])
     self.assertAllClose(iou_output, exp_output)
 
+  def test_l1(self):
+    def graph_fn():
+      corners1 = tf.constant([[4.0, 3.0, 7.0, 5.0], [5.0, 6.0, 10.0, 7.0]])
+      corners2 = tf.constant([[3.0, 4.0, 6.0, 8.0], [14.0, 14.0, 15.0, 15.0],
+                              [0.0, 0.0, 20.0, 20.0]])
+      boxes1 = box_list.BoxList(corners1)
+      boxes2 = box_list.BoxList(corners2)
+      l1 = box_list_ops.l1(boxes1, boxes2)
+      return l1
+    exp_output = [[5.0, 22.5, 45.5], [8.5, 19.0, 40.0]]
+    l1_output = self.execute(graph_fn, [])
+    self.assertAllClose(l1_output, exp_output)
+
+  def test_giou(self):
+    def graph_fn():
+      corners1 = tf.constant([[5.0, 7.0, 7.0, 9.0]])
+      corners2 = tf.constant([[5.0, 7.0, 7.0, 9.0], [5.0, 11.0, 7.0, 13.0]])
+      boxes1 = box_list.BoxList(corners1)
+      boxes2 = box_list.BoxList(corners2)
+      giou = box_list_ops.giou_loss(boxes1, boxes2)
+      return giou
+    exp_output = [[0.0, 4.0 / 3.0]]
+    giou_output = self.execute(graph_fn, [])
+    self.assertAllClose(giou_output, exp_output)
+
   def test_matched_iou(self):
     def graph_fn():
       corners1 = tf.constant([[4.0, 3.0, 7.0, 5.0], [5.0, 6.0, 10.0, 7.0]])

research/object_detection/core/losses.py

@@ -36,7 +36,8 @@ import tensorflow.compat.v1 as tf
 from object_detection.core import box_list
 from object_detection.core import box_list_ops
 from object_detection.utils import ops
-
+from object_detection.box_coders import detr_box_coder
+import tensorflow_addons as tfa
 
 class Loss(six.with_metaclass(abc.ABCMeta, object)):
   """Abstract base class for loss functions."""
@@ -180,7 +181,6 @@ class WeightedSmoothL1LocalizationLoss(Loss):
         reduction=tf.losses.Reduction.NONE
     ), axis=2)
 
-
 class WeightedIOULocalizationLoss(Loss):
   """IOU localization loss function.
 
@@ -203,12 +203,44 @@ class WeightedIOULocalizationLoss(Loss):
       loss: a float tensor of shape [batch_size, num_anchors] tensor
         representing the value of the loss function.
     """
-    predicted_boxes = box_list.BoxList(tf.reshape(prediction_tensor, [-1, 4]))
-    target_boxes = box_list.BoxList(tf.reshape(target_tensor, [-1, 4]))
+    predicted_boxes = prediction_tensor# box_list.BoxList(tf.reshape(prediction_tensor, [-1, 4]))
+    target_boxes = target_tensor #box_list.BoxList(tf.reshape(target_tensor, [-1, 4]))
     per_anchor_iou_loss = 1.0 - box_list_ops.matched_iou(predicted_boxes,
                                                          target_boxes)
+    print("Weights", weights)
     return tf.reshape(weights, [-1]) * per_anchor_iou_loss
 
+class WeightedGIOULocalizationLoss(Loss):
+  """IOU localization loss function.
+
+  Sums the IOU for corresponding pairs of predicted/groundtruth boxes
+  and for each pair assign a loss of 1 - IOU.  We then compute a weighted
+  sum over all pairs which is returned as the total loss.
+  """
+
+  def _compute_loss(self, prediction_tensor, target_tensor, weights):
+    """Compute loss function.
+
+    Args:
+      prediction_tensor: A float tensor of shape [batch_size, num_anchors, 4]
+        representing the decoded predicted boxes
+      target_tensor: A float tensor of shape [batch_size, num_anchors, 4]
+        representing the decoded target boxes
+      weights: a float tensor of shape [batch_size, num_anchors]
+
+    Returns:
+      loss: a float tensor of shape [batch_size, num_anchors] tensor
+        representing the value of the loss function.
+    """
+    predicted_boxes = prediction_tensor# box_list.BoxList(tf.reshape(prediction_tensor, [-1, 4]))
+    target_boxes = target_tensor #box_list.BoxList(tf.reshape(target_tensor, [-1, 4]))
+    #loss_function = tfa.losses.GIouLoss()
+    per_anchor_iou_loss = tfa.losses.giou_loss(predicted_boxes,target_boxes) 
+                          #                               #1.0 - box_list_ops.matched_iou(predicted_boxes,
+                          #                               target_boxes)
+    #print("Weights", weights)
+    return tf.reshape(weights, [-1]) * per_anchor_iou_loss # * per_anchor_iou_loss
+
 
 class WeightedSigmoidClassificationLoss(Loss):
   """Sigmoid cross entropy classification loss function."""