Skip to content

make CTD GenerativeSampler visibility-aware #3305

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
deruyter92 merged 2 commits into from
May 12, 2026
Merged

make CTD GenerativeSampler visibility-aware #3305

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9663286
make GenerativeSampler visibility-aware
deruyter92 May 4, 2026
d48f369
Merge branch 'main' into jaap/update_ctd_generative_sampler
C-Achard May 12, 2026
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
8 changes: 8 additions & 0 deletions deeplabcut/pose_estimation_pytorch/data/generative_sampling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,6 +173,14 @@ def __call__(

N = 500 # TODO: do not know how this is set
for j in range(self.num_keypoints):
# Skip unlabeled / invisible GT joints. Synthesizing a conditional
# keypoint for them creates a spurious cue and biases CTD training.
# Previously, this was prevented implicitly by NaN propagation; now
# we make the contract explicit. (Required since PR #2995)
if keypoints[j, 2] <= 0:
synth_joints[j] = 0 # (x, y, vis) = (0, 0, 0)
continue

# source keypoint position candidates to generate error on that (gt, swap, inv, swap+inv)
coord_list = []
# on top of gt
Expand Down
Loading