Skip to content

T3A — Pose / keypoints — camera & studio shift

Paper evidence: main.pdf · Block findings

Lemma: D3 · Stack: pytorch Nuisance key: augmentation

Production change: Camera, lighting, viewpoint change; same keypoint indices.

Notebook (Run All, built-in demo): t03a-pose-gradient.ipynb

pip install matching-pmh torch
# Open the notebook and Run All

What this task achieved (headline)

E1_aniso subspace PMH on COCO pose: 54.49% clean PCK@0.05 (+22.4 pp vs baseline 32.07%).

baseline E1_aniso
32.07% PCK 54.49% PCK
35.21% @ occ 0.40

Subtasks (paper)

Calibrate occlusion subspace W

Gradient-SVD on COCO pose features.

Train baseline / E1 / E1_aniso / VAT

E1_aniso +22.4 pp PCK vs baseline.

Robustness + embedding eval

Occlusion stress + drift metrics.

Run with matching-pmh

from pmh import PMHTrainer, evaluate_robust_fit
# nuisance="augmentation"

Do not use PMH when

Different skeleton or keypoint definitions at deploy.

Replace demo data with yours

Swap demo loaders for your train_loader, source_batches, target_batches, and deploy holdout. Hook the backbone before your task head.

← All 13 tasks · Quickstart