Skip to content

Properly handle empty assemblies #1641

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
MMathisLab merged 1 commit into from
Jan 18, 2022
Merged

Properly handle empty assemblies #1641

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion deeplabcut/pose_estimation_tensorflow/lib/inferenceutils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,6 +314,12 @@ def calc_assembly_mahalanobis_dist(
if self._kde is None:
raise ValueError("Assembler should be calibrated first with training data.")

if not len(assembly):
# Distance is undefined if the assembly is empty
if return_proba:
return np.inf, 0
return np.inf

dists = assembly.calc_pairwise_distances() - self._kde.mean
mask = np.isnan(dists)
if mask.any() and nan_policy == "little":
Expand Down Expand Up @@ -374,6 +380,8 @@ def extract_best_links(self, joints_dict, costs, trees=None):
dets_t = joints_dict.get(t, None)
if dets_s is None or dets_t is None:
continue
if ind not in costs:
continue
lengths = costs[ind]["distance"]
if np.isinf(lengths).all():
continue
Expand Down Expand Up @@ -705,7 +713,7 @@ def _assemble(self, data_dict, ind_frame):
if joint.idx not in assembled and np.isfinite(joint.confidence)
)
for assembly in assemblies[::-1]:
if 0 < assembly.n_links < self.min_n_links:
if 0 <= assembly.n_links < self.min_n_links:
for link in assembly._links:
discarded.update((link.j1, link.j2))
assemblies.remove(assembly)
Expand Down