data-datasets
Register and validate Detectron2 datasets, metadata, COCO helpers, mappers, and data loaders.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- data-datasets
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Register and validate Detectron2 datasets, metadata, COCO helpers, mappers, and data loaders.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/data-loading.md7.5 KB
- 参考文档references/dataset-format.md8.1 KB
- 参考文档references/troubleshooting.md6.9 KB
- 脚本scripts/validate_dataset_dicts.py22.1 KB
- 脚本scripts/validate_dataset_registration.py9.8 KB
03 SKILL.md 正文
Detectron2 Data & Datasets
Use this sub-skill when a task is about getting data into Detectron2: registering custom datasets, checking Detectron2-standard dataset dicts, setting metadata, using COCO helpers, designing mappers, or building train/test data loaders.
Read First
references/dataset-format.mdforDatasetCatalog,MetadataCatalog, standard dataset dict fields, COCO registration, metadata keys, and dataset-root conventions.references/data-loading.mdforDatasetMapper, custom mapper patterns, augmentations, andbuild_detection_train_loader/build_detection_test_loaderusage.references/troubleshooting.mdfor common registration, annotation, metadata, class-count, COCO path, and mapper failures.
Common Workflows
- Register a dataset by name with
DatasetCatalog.register(name, callable); the callable must take no arguments and return the same ordered records on every call. - Attach shared dataset facts through
MetadataCatalog.get(name).set(...), especially class names and evaluator/COCO metadata. - Validate records with
scripts/validate_dataset_dicts.pybefore wiring them into a loader or config. - Use
register_coco_instancesorload_coco_jsonwhen annotations are COCO instance/keypoint style. - Pass the registered names into
cfg.DATASETS.TRAIN/cfg.DATASETS.TEST, and route class-count or solver changes to the config/training sub-skills. - Customize loading through
DatasetMapper(..., augmentations=...)or a custom mapper passed tobuild_detection_train_loader(..., mapper=...).
Bundled Helpers
scripts/validate_dataset_dicts.py: validates a JSON list of Detectron2-standard dataset dicts and reports actionable schema/data-shape errors.scripts/validate_dataset_registration.py: smoke-checks catalog registration and metadata behavior with a synthetic dataset, and optionally validates a user JSON fixture.
Boundaries
- This sub-skill owns dataset registration, metadata, dataset dict structure, COCO helpers, mapper design, loader construction, dataset roots, and annotation/loader troubleshooting.
- Use the training/evaluation sub-skill for
DefaultTrainer, launch commands, evaluators, class-count config changes, and metrics. - Use the inference/visualization sub-skill for rendering predictions or dataset visualizations.
- Use extension-projects guidance for project-specific dataset converters or custom task heads.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).