Add MPI-INF-3DHP test evaluation workflow

[xiu-cs]

Summary

This PR adds a self-contained MPI-INF-3DHP test evaluation workflow for monocular 3D pose lifting models, including a processed-test-set loader, inference entrypoint, shell launcher, and metric reporting for MPJPE, P-MPJPE, PCK, and AUC.

It also documents how to prepare the official MPI-INF-3DHP test annotations into the expected .npz format, how to configure packaged or local model definitions, and how to use either auto-downloaded or local human FMPose3D weights.

Changes

• Add 3dhp_test/infer_3dhp.py for running 3DHP test-set inference with multi-hypothesis flow-matching sampling and RPEA aggregation.
• Add 3DHP-specific dataset, camera, metric, and utility helpers under 3dhp_test/lib/.
• Add 3dhp_test/test_3dhp.sh with configurable model_type, model_weights_path, model_path, subjects, batch size, GPU, and sampling settings.
• Add 3dhp_test/dataset/prepare_3dhp_test_npz.py to convert official MPI-INF-3DHP TS*/annot_data.mat files into data_test_3dhp.npz.
• Document dataset preparation, model/weights configuration, and P-STMO preprocessing acknowledgement in 3dhp_test/README.md.
• Add ignore rules so generated .npz files, results, caches, and local checkpoints are not committed while 3dhp_test/lib/*.py remains trackable.
• Add tests covering registry model loading, per-sample camera tensor selection for mixed-subject batches, and use of the shared FMPose3D weight resolver.

[Copilot]

Pull request overview

Adds a new, self-contained workflow under 3dhp_test/ to run MPI-INF-3DHP test-set inference for monocular 3D pose lifting models and report standard 3DHP metrics, plus documentation and ignore rules to keep generated artifacts out of the repo.

Changes:

• Introduces a 3DHP test inference entrypoint (infer_3dhp.py) with multi-hypothesis sampling, RPEA aggregation, and metric reporting (MPJPE / P-MPJPE / PCK / AUC).
• Adds 3DHP-specific helpers for dataset loading, camera normalization/projection, and metric computation under 3dhp_test/lib/.
• Adds dataset preparation script + user documentation + gitignore updates for generated .npz and evaluation outputs.

Reviewed changes

Copilot reviewed 9 out of 11 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
3dhp_test/test_3dhp.sh	Shell launcher for running the 3DHP test inference with configurable model/weights/sampling parameters.
3dhp_test/README.md	Documents model/weights configuration and how to convert official MPI-INF-3DHP test annotations into the expected .npz.
3dhp_test/lib/utils.py	Minimal utilities for accumulating/formatting error metrics and defining the 3DHP action list.
3dhp_test/lib/eval_cal.py	Implements MPJPE, P-MPJPE, PCK, and AUC computations and per-action aggregation/printing.
3dhp_test/lib/dataset_3dhp.py	Loads processed test .npz, applies valid mask + joint mapping + normalization + optional flip augmentation.
3dhp_test/lib/camera.py	3DHP-specific screen normalization, projection model, and subject-dependent intrinsics/resolution.
3dhp_test/lib/__init__.py	Marks 3dhp_test/lib as a package.
3dhp_test/infer_3dhp.py	Main inference/evaluation script that loads a model, resolves weights, runs sampling + aggregation, and writes metrics JSON/logs.
3dhp_test/dataset/prepare_3dhp_test_npz.py	Converts official TS*/annot_data.mat files into a compressed .npz consumed by the evaluator.
3dhp_test/.gitignore	Prevents committing generated results, local checkpoints, caches, and dataset .npz files in 3dhp_test/.
.gitignore	Ensures 3dhp_test/lib/*.py remains trackable despite the repository’s broader lib/ ignore rule; normalizes a predictions ignore entry.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 10 out of 12 changed files in this pull request and generated 4 comments.