def test_symmetry():
s1 = Skeleton("s1")
s1.add_nodes(["1", "2", "3", "4", "5", "6"])
s1.add_edge("1", "2")
s1.add_edge("3", "4")
s1.add_edge("5", "6")
s1.add_symmetry("1", "5")
s1.add_symmetry("3", "6")
assert s1.get_symmetry("1").name == "5"
assert s1.get_symmetry("5").name == "1"
assert s1.get_symmetry("3").name == "6"
# Cannot add more than one symmetry to a node
with pytest.raises(ValueError):
s1.add_symmetry("1", "6")
with pytest.raises(ValueError):
s1.add_symmetry("6", "1")
s1.delete_symmetry("1", "5")
assert s1.get_symmetry("1") is None
with pytest.raises(ValueError):
s1.delete_symmetry("1", "5")
def test_eq():
s1 = Skeleton("s1")
s1.add_nodes(["1", "2", "3", "4", "5", "6"])
s1.add_edge("1", "2")
s1.add_edge("3", "4")
s1.add_edge("5", "6")
s1.add_symmetry("3", "6")
# Make a copy check that they are equal
s2 = copy.deepcopy(s1)
assert s1.matches(s2)
# Add an edge, check that they are not equal
s2 = copy.deepcopy(s1)
s2.add_edge("5", "1")
assert not s1.matches(s2)
# Add a symmetry edge, not equal
s2 = copy.deepcopy(s1)
s2.add_symmetry("5", "1")
assert not s1.matches(s2)
# Delete a node
s2 = copy.deepcopy(s1)
s2.delete_node("5")
assert not s1.matches(s2)
# Delete and edge, not equal
s2 = copy.deepcopy(s1)
s2.delete_edge("1", "2")
assert not s1.matches(s2)
def test_symmetry():
s1 = Skeleton("s1")
s1.add_nodes(["1", "2", "3", "4", "5", "6"])
s1.add_edge("1", "2")
s1.add_edge("3", "4")
s1.add_edge("5", "6")
s1.add_symmetry("1", "5")
s1.add_symmetry("3", "6")
assert (s1.nodes[0], s1.nodes[4]) in s1.symmetries
assert (s1.nodes[2], s1.nodes[5]) in s1.symmetries
assert len(s1.symmetries) == 2
assert (0, 4) in s1.symmetric_inds
assert (2, 5) in s1.symmetric_inds
assert len(s1.symmetric_inds) == 2
assert s1.get_symmetry("1").name == "5"
assert s1.get_symmetry("5").name == "1"
assert s1.get_symmetry("3").name == "6"
# Cannot add more than one symmetry to a node
with pytest.raises(ValueError):
s1.add_symmetry("1", "6")
with pytest.raises(ValueError):
s1.add_symmetry("6", "1")
s1.delete_symmetry("1", "5")
assert s1.get_symmetry("1") is None
with pytest.raises(ValueError):
s1.delete_symmetry("1", "5")
def stickman():
# Make a skeleton with a space in its name to test things.
stickman = Skeleton("Stick man")
stickman.add_nodes(
["head", "neck", "body", "right-arm", "left-arm", "right-leg", "left-leg"]
)
stickman.add_edge("neck", "head")
stickman.add_edge("body", "neck")
stickman.add_edge("body", "right-arm")
stickman.add_edge("body", "left-arm")
stickman.add_edge("body", "right-leg")
stickman.add_edge("body", "left-leg")
stickman.add_symmetry(node1="left-arm", node2="right-arm")
stickman.add_symmetry(node1="left-leg", node2="right-leg")
return stickman
def skeleton():
# Create a simple skeleton object
skeleton = Skeleton("Fly")
skeleton.add_node("head")
skeleton.add_node("thorax")
skeleton.add_node("abdomen")
skeleton.add_node("left-wing")
skeleton.add_node("right-wing")
skeleton.add_edge(source="head", destination="thorax")
skeleton.add_edge(source="thorax", destination="abdomen")
skeleton.add_edge(source="thorax", destination="left-wing")
skeleton.add_edge(source="thorax", destination="right-wing")
skeleton.add_symmetry(node1="left-wing", node2="right-wing")
return skeleton
def test_skeleton_node_name_change():
"""
Test that and instance is not broken after a node on the
skeleton has its name changed.
"""
s = Skeleton("Test")
s.add_nodes(["a", "b", "c", "d", "e"])
s.add_edge("a", "b")
instance = Instance(s)
instance["a"] = Point(1, 2)
instance["b"] = Point(3, 4)
# Rename the node
s.relabel_nodes({"a": "A"})
# Reference to the old node name should raise a KeyError
with pytest.raises(KeyError):
instance["a"].x = 2
# Make sure the A now references the same point on the instance
assert instance["A"] == Point(1, 2)
assert instance["b"] == Point(3, 4)
def test_arborescence():
skeleton = Skeleton()
skeleton.add_node("a")
skeleton.add_node("b")
skeleton.add_node("c")
# linear: a -> b -> c
skeleton.add_edge("a", "b")
skeleton.add_edge("b", "c")
assert skeleton.is_arborescence
skeleton = Skeleton()
skeleton.add_node("a")
skeleton.add_node("b")
skeleton.add_node("c")
# two branches from a: a -> b and a -> c
skeleton.add_edge("a", "b")
skeleton.add_edge("a", "c")
assert skeleton.is_arborescence
skeleton = Skeleton()
skeleton.add_node("a")
skeleton.add_node("b")
skeleton.add_node("c")
# no edges so too many roots
assert not skeleton.is_arborescence
assert sorted((n.name for n in skeleton.root_nodes)) == ["a", "b", "c"]
# still too many roots: a and c
skeleton.add_edge("a", "b")
assert not skeleton.is_arborescence
assert sorted((n.name for n in skeleton.root_nodes)) == ["a", "c"]
skeleton = Skeleton()
skeleton.add_node("a")
skeleton.add_node("b")
skeleton.add_node("c")
# cycle
skeleton.add_edge("a", "b")
skeleton.add_edge("b", "c")
skeleton.add_edge("c", "a")
assert not skeleton.is_arborescence
assert len(skeleton.cycles) == 1
assert len(skeleton.root_nodes) == 0
skeleton = Skeleton()
skeleton.add_node("a")
skeleton.add_node("b")
skeleton.add_node("c")
skeleton.add_node("d")
# diamond, too many sources leading to d
skeleton.add_edge("a", "b")
skeleton.add_edge("a", "c")
skeleton.add_edge("b", "d")
skeleton.add_edge("c", "d")
assert not skeleton.is_arborescence
assert len(skeleton.cycles) == 0
assert len(skeleton.root_nodes) == 1
assert len(skeleton.in_degree_over_one) == 1
def load_predicted_labels_json_old(
data_path: str,
parsed_json: dict = None,
adjust_matlab_indexing: bool = True,
fix_rel_paths: bool = True,
) -> List[LabeledFrame]:
"""
Load predicted instances from Talmo's old JSON format.
Args:
data_path: The path to the JSON file.
parsed_json: The parsed json if already loaded, so we can save
some time if already parsed.
adjust_matlab_indexing: Whether to adjust indexing from MATLAB.
fix_rel_paths: Whether to fix paths to videos to absolute paths.
Returns:
List of :class:`LabeledFrame` objects.
"""
if parsed_json is None:
data = json.loads(open(data_path).read())
else:
data = parsed_json
videos = pd.DataFrame(data["videos"])
predicted_instances = pd.DataFrame(data["predicted_instances"])
predicted_points = pd.DataFrame(data["predicted_points"])
if adjust_matlab_indexing:
predicted_instances.frameIdx -= 1
predicted_points.frameIdx -= 1
predicted_points.node -= 1
predicted_points.x -= 1
predicted_points.y -= 1
skeleton = Skeleton()
skeleton.add_nodes(data["skeleton"]["nodeNames"])
edges = data["skeleton"]["edges"]
if adjust_matlab_indexing:
edges = np.array(edges) - 1
for (src_idx, dst_idx) in edges:
skeleton.add_edge(
data["skeleton"]["nodeNames"][src_idx],
data["skeleton"]["nodeNames"][dst_idx],
)
if fix_rel_paths:
for i, row in videos.iterrows():
p = row.filepath
if not os.path.exists(p):
p = os.path.join(os.path.dirname(data_path), p)
if os.path.exists(p):
videos.at[i, "filepath"] = p
# Make the video objects
video_objects = {}
for i, row in videos.iterrows():
if videos.at[i, "format"] == "media":
vid = Video.from_media(videos.at[i, "filepath"])
else:
vid = Video.from_hdf5(
filename=videos.at[i, "filepath"], dataset=videos.at[i, "dataset"]
)
video_objects[videos.at[i, "id"]] = vid
track_ids = predicted_instances["trackId"].values
unique_track_ids = np.unique(track_ids)
spawned_on = {
track_id: predicted_instances.loc[predicted_instances["trackId"] == track_id][
"frameIdx"
].values[0]
for track_id in unique_track_ids
}
tracks = {
i: Track(name=str(i), spawned_on=spawned_on[i])
for i in np.unique(predicted_instances["trackId"].values).tolist()
}
# A function to get all the instances for a particular video frame
def get_frame_predicted_instances(video_id, frame_idx):
points = predicted_points
is_in_frame = (points["videoId"] == video_id) & (
points["frameIdx"] == frame_idx
)
if not is_in_frame.any():
return []
instances = []
frame_instance_ids = np.unique(points["instanceId"][is_in_frame])
for i, instance_id in enumerate(frame_instance_ids):
is_instance = is_in_frame & (points["instanceId"] == instance_id)
track_id = predicted_instances.loc[
predicted_instances["id"] == instance_id
]["trackId"].values[0]
match_score = predicted_instances.loc[
predicted_instances["id"] == instance_id
]["matching_score"].values[0]
track_score = predicted_instances.loc[
predicted_instances["id"] == instance_id
]["tracking_score"].values[0]
instance_points = {
data["skeleton"]["nodeNames"][n]: PredictedPoint(
x, y, visible=v, score=confidence
)
for x, y, n, v, confidence in zip(
*[
points[k][is_instance]
for k in ["x", "y", "node", "visible", "confidence"]
]
)
}
instance = PredictedInstance(
skeleton=skeleton,
points=instance_points,
track=tracks[track_id],
score=match_score,
)
instances.append(instance)
return instances
# Get the unique labeled frames and construct a list of LabeledFrame objects for them.
frame_keys = list(
{
(videoId, frameIdx)
for videoId, frameIdx in zip(
predicted_points["videoId"], predicted_points["frameIdx"]
)
}
)
frame_keys.sort()
labels = []
for videoId, frameIdx in frame_keys:
label = LabeledFrame(
video=video_objects[videoId],
frame_idx=frameIdx,
instances=get_frame_predicted_instances(videoId, frameIdx),
)
labels.append(label)
return labels
