def make_labeled_frames_from_generator(self, generator, data_provider):
grouped_generator = group_examples_iter(generator)
skeleton = self.bottomup_config.data.labels.skeletons[0]
def make_lfs(video_ind, frame_ind, frame_examples):
return make_grouped_labeled_frame(
video_ind=video_ind,
frame_ind=frame_ind,
frame_examples=frame_examples,
videos=data_provider.videos,
skeleton=skeleton,
image_key="image",
points_key="predicted_instances",
point_confidences_key="predicted_peak_scores",
instance_score_key="predicted_instance_scores",
tracker=self.tracker,
)
predicted_frames = []
for (video_ind, frame_ind), grouped_examples in grouped_generator:
predicted_frames.extend(make_lfs(video_ind, frame_ind, grouped_examples))
if self.tracker:
self.tracker.final_pass(predicted_frames)
return predicted_frames
def test_group_iterator():
examples = make_examples()
# Use iterator to build grouped dict
grouped = dict()
for key, val in group_examples_iter(examples):
grouped[key] = val
check_grouped_examples(grouped)
def make_labeled_frames_from_generator(self, generator, data_provider):
grouped_generator = group_examples_iter(generator)
skeleton = self.confmap_config.data.labels.skeletons[0]
def make_lfs(video_ind, frame_ind, frame_examples):
return make_grouped_labeled_frame(
video_ind=video_ind,
frame_ind=frame_ind,
frame_examples=frame_examples,
videos=data_provider.videos,
skeleton=skeleton,
points_key="predicted_instance",
point_confidences_key="predicted_instance_confidences",
)
predicted_frames = []
for (video_ind, frame_ind), grouped_examples in grouped_generator:
predicted_frames.extend(make_lfs(video_ind, frame_ind, grouped_examples))
return predicted_frames