fix suggestions in utils

research/object_detection/dataset_tools/seq_example_util.py

@@ -200,6 +200,8 @@ def make_sequence_example(dataset_name,
     detection_scores: (Optional) A list (with num_frames_elements) of
       [num_boxes_i] numpy float32 arrays holding predicted object scores for
       each frame.
+    use_strs_for_source_id: (Optional) Whether to write the source IDs as strings
+      rather than byte lists of characters.
 
   Returns:
     A tf.train.SequenceExample.

research/object_detection/dataset_tools/seq_example_util_test.py

@@ -104,12 +104,12 @@ class SeqExampleUtilTest(tf.test.TestCase):
         source_ids)
 
   def test_make_labeled_example(self):
-    num_frames = 2
+    num_frames = 3
     image_height = 100
     image_width = 200
     dataset_name = b'unlabeled_dataset'
     video_id = b'video_000'
-    labels = [b'dog', b'cat']
+    labels = [b'dog', b'cat', b'wolf']
     images = tf.cast(tf.random.uniform(
         [num_frames, image_height, image_width, 3],
         maxval=256,
@@ -117,8 +117,8 @@ class SeqExampleUtilTest(tf.test.TestCase):
     images_list = tf.unstack(images, axis=0)
     encoded_images_list = [tf.io.encode_jpeg(image) for image in images_list]
     encoded_images = self.materialize_tensors(encoded_images_list)
-    timestamps = [100000, 110000]
-    is_annotated = [1, 0]
+    timestamps = [100000, 110000, 120000]
+    is_annotated = [1, 0, 1]
     bboxes = [
         np.array([[0., 0., 0., 0.],
                   [0., 0., 1., 1.]], dtype=np.float32),
@@ -127,6 +127,7 @@ class SeqExampleUtilTest(tf.test.TestCase):
     ]
     label_strings = [
         np.array(labels),
+        np.array([]),
         np.array([])
     ]