L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Données et recherche · SKILL.md

dataset-preparation

dataset-preparation

Validate nuScenes + CAN-bus layout and the generated BEVFormer temporal annotation files.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: dataset-preparation description: "Validate nuScenes + CAN-bus layout and the generated BEVFormer temporal annotation files." disable-model-invocation: true metadata: disco-role: operating license: Apache 2.0

Dataset Preparation

Use this sub-skill when a task depends on the nuScenes raw tree, CAN-bus expansion, or the temporal annotation files consumed by BEVFormer and BEVFormerV2.

Covers

  • nuScenes folder layout checks for the raw samples/, sweeps/, maps/, and version folders.
  • CAN-bus expansion placement and the pose metadata it feeds.
  • Generated nuscenes_infos_temporal_train.pkl, nuscenes_infos_temporal_val.pkl, and nuscenes_infos_temporal_test.pkl.
  • data_root / ann_file wiring for camera-only BEVFormer configs.
  • Metadata keys emitted by CustomNuScenesDataset and CustomNuScenesDatasetV2.

Routes elsewhere

  • Model, backbone, transformer, or other config knobs belong in installation-and-configs.
  • Distributed training, evaluation, checkpoints, or FP16 launch commands belong in training-and-evaluation.
  • Log analysis, benchmark, or visualization tasks belong in analysis-and-utilities.

Start here

  1. Read nuScenes layout notes.
  2. Read data format notes.
  3. Read troubleshooting notes.
  4. Run the bundled layout checker.
Skills enregistrées