diff --git a/source/utils.py b/source/utils.py
index 1d02602e27ab213aeb8802ec15b2fd6aa20b56a7..b85655493adedc2d9488be208968418d18bf2a7e 100644
--- a/source/utils.py
+++ b/source/utils.py
@@ -23,11 +23,21 @@ from tqdm import tqdm_notebook as tqdm
# QUALITY CONTROL AND PREPROCESSING #
-def likelihood_qc(dframe, threshold=0.9):
- """Returns only rows where all lilelihoods are above a specified threshold"""
+def likelihood_qc(dframe: pd.DataFrame, threshold: float = 0.9) -> np.array:
+ """Returns a DataFrame filtered dataframe, keeping only the rows entirely above the threshold.
+
+ Parameters:
+ - dframe (pandas.DataFrame): DeepLabCut output, with positions over time and associated likelihhod
+ - threshold (float): minimum acceptable confidence
+
+ Returns:
+ - filt_mask (np.array): mask on the rows of dframe"""
+
Likes = np.array([dframe[i]["likelihood"] for i in list(dframe.columns.levels[0])])
Likes = np.nan_to_num(Likes, nan=1.0)
- return np.all(Likes > threshold, axis=0)
+ filt_mask = np.all(Likes > threshold, axis=0)
+
+ return filt_mask
def bp2polar(tab: pd.DataFrame) -> pd.DataFrame:
@@ -244,10 +254,9 @@ def rolling_window(a: np.array, window_size: int, window_step: int) -> np.array:
def smooth_mult_trajectory(series: np.array, alpha: float = 0.15) -> np.array:
"""Returns a smooths a trajectory using exponentially weighted averages
- Parameters:
- - series (np.array): 2D trajectory array with N (instances) * m (features)
- - alpha (float): 0 <= alpha <= 1; indicates the weight assigned to the current observation.
- higher (alpha~1) indicates less smoothing; lower indicates more (alpha~0)
+ Parameters: - series (numpyp.array): 1D trajectory array with N (instances) - alpha (float): 0 <= alpha <= 1;
+ indicates the inverse weight assigned to previous observations. Higher (alpha~1) indicates less smoothing; lower
+ indicates more (alpha~0)
Returns:
- smoothed_series (np.array): smoothed version of the input, with equal shape"""
@@ -260,7 +269,8 @@ def smooth_mult_trajectory(series: np.array, alpha: float = 0.15) -> np.array:
return smoothed_series
- ##### IMAGE/VIDEO PROCESSING FUNCTIONS #####
+
+# IMAGE/VIDEO PROCESSING FUNCTIONS #
def index_frames(video_list, sample=False, index=0, pkl=False):
@@ -286,7 +296,8 @@ def index_frames(video_list, sample=False, index=0, pkl=False):
return True
- ##### BEHAVIOUR RECOGNITION FUNCTIONS #####
+
+# BEHAVIOUR RECOGNITION FUNCTIONS #
# Nose to Nose contact
diff --git a/test_deepof/test_utils.py b/test_deepof/test_utils.py
index 69459ccba27f5ca9424fe02ef23dadddefdfa8fa..012cd816fac2aa116924211f0ef36c6c7ea0e950 100644
--- a/test_deepof/test_utils.py
+++ b/test_deepof/test_utils.py
@@ -4,14 +4,54 @@ import sys
sys.path.append("../")
from hypothesis import given
+from hypothesis import settings
from hypothesis import strategies as st
from hypothesis.extra.numpy import arrays
from hypothesis.extra.pandas import range_indexes, columns, data_frames
-from itertools import combinations
from scipy.spatial import distance
from source.utils import *
+@given(
+ mult=st.integers(min_value=1, max_value=10),
+ dframe=data_frames(
+ index=range_indexes(min_size=1),
+ columns=columns(["X", "y", "likelihood"], dtype=float),
+ rows=st.tuples(
+ st.floats(
+ min_value=0, max_value=1000, allow_nan=False, allow_infinity=False
+ ),
+ st.floats(
+ min_value=0, max_value=1000, allow_nan=False, allow_infinity=False
+ ),
+ st.floats(
+ min_value=0.01, max_value=1.0, allow_nan=False, allow_infinity=False
+ ),
+ ),
+ ),
+ threshold=st.data(),
+)
+def test_likelihood_qc(mult, dframe, threshold):
+ thresh1 = threshold.draw(st.floats(min_value=0.1, max_value=1.0, allow_nan=False))
+ thresh2 = threshold.draw(
+ st.floats(min_value=thresh1, max_value=1.0, allow_nan=False)
+ )
+
+ dframe = pd.concat([dframe] * mult, axis=0)
+ idx = pd.MultiIndex.from_product(
+ [list(dframe.columns[: len(dframe.columns) // 3]), ["X", "y", "likelihood"]],
+ names=["bodyparts", "coords"],
+ )
+ dframe.columns = idx
+
+ filt1 = likelihood_qc(dframe, thresh1)
+ filt2 = likelihood_qc(dframe, thresh2)
+
+ assert np.sum(filt1) <= dframe.shape[0]
+ assert np.sum(filt2) <= dframe.shape[0]
+ assert np.sum(filt1) >= np.sum(filt2)
+
+
@given(
tab=data_frames(
index=range_indexes(min_size=1),
@@ -119,7 +159,6 @@ def test_bpart_distance(cordarray):
),
)
def test_angle(abc):
-
a, b, c = abc
angles = []
@@ -164,7 +203,6 @@ def test_angle_trio(array):
)
)
def test_rotate(p):
-
assert np.allclose(rotate(p, 2 * np.pi), p)
assert np.allclose(rotate(p, np.pi), -p)
@@ -195,7 +233,10 @@ def test_align_trajectories(data):
@given(a=arrays(dtype=bool, shape=st.tuples(st.integers(min_value=3, max_value=1000))))
def test_smooth_boolean_array(a):
smooth = smooth_boolean_array(a)
- trans = lambda x: sum([i + 1 != i for i in range(x.shape[0] - 1)])
+
+ def trans(x):
+ return sum([i + 1 != i for i in range(x.shape[0] - 1)])
+
assert trans(a) >= trans(smooth)
@@ -222,3 +263,37 @@ def test_rolling_window(a, window):
assert len(rolled_shape) == len(a.shape) + 1
assert rolled_shape[1] == window_size
+
+
+@settings(deadline=None)
+@given(
+ alpha=st.data(),
+ series=arrays(
+ dtype=float,
+ shape=st.tuples(st.integers(min_value=10, max_value=1000),),
+ elements=st.floats(
+ min_value=1.0, max_value=1.0, allow_nan=False, allow_infinity=False
+ ),
+ ),
+)
+def test_smooth_mult_trajectory(alpha, series):
+ alpha1 = alpha.draw(
+ st.floats(min_value=0.1, max_value=1.0, allow_nan=False, allow_infinity=False)
+ )
+ alpha2 = alpha.draw(
+ st.floats(
+ min_value=alpha1, max_value=1.0, allow_nan=False, allow_infinity=False
+ )
+ )
+
+ series *= +np.random.normal(0, 1, len(series))
+
+ smoothed1 = smooth_mult_trajectory(series, alpha1)
+ smoothed2 = smooth_mult_trajectory(series, alpha2)
+
+ def autocorr(x, t=1):
+ return np.corrcoef(np.array([x[:-t], x[t:]]))[0, 1]
+
+ assert autocorr(smoothed1) >= autocorr(series)
+ assert autocorr(smoothed2) >= autocorr(series)
+ assert autocorr(smoothed2) <= autocorr(smoothed1)