tracking
Parse KITTI tracking labels with pykitti==0.3.1 and assess its incomplete tracking loader conservatively.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- tracking
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Parse KITTI tracking labels with pykitti==0.3.1 and assess its incomplete tracking loader conservatively.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/api-reference.md7.8 KB
- 参考文档references/data-format.md5.5 KB
- 参考文档references/troubleshooting.md5.4 KB
- 参考文档references/workflows.md5.7 KB
- 脚本scripts/labels_fixture_smoke.py5.0 KB
03 SKILL.md 正文
KITTI tracking labels
Use this sub-skill primarily for the two usable label utilities in
pykitti==0.3.1: KittiTrackingLabels and to_array_list. They parse local
KITTI tracking/detection records into per-frame arrays; they do not acquire a
dataset and do not repair malformed data.
Route and boundary
- Use
KittiTrackingLabels(path_or_df, bbox_with_size=True, remove_dontcare=True, split_on_reappear=True)for a label/detection text file or a prepared pandasDataFrame. - Use
to_array_list(df, length=None, by_id=True)for a frame-indexedDataFrame. Its index is the frame number, not a column. - Treat
pykitti.tracking(base_path, sequence, **kwargs)as an incomplete, legacy sensor-loader skeleton, not as a verified tracking dataset loader. It has no initialized timestamps, prints a file count, and its frame sub-selection references missingcam0_files/cam1_filesattributes. See API reference and troubleshooting. - Use raw-data for raw drives and odometry for odometry sequences. Read the shared root troubleshooting guide for installation and package-wide issues.
Minimal safe route
from pykitti.tracking import KittiTrackingLabels, to_array_list
labels = KittiTrackingLabels(
"/data/kitti/tracking/training/label_02/0000.txt",
remove_dontcare=True,
split_on_reappear=True,
)
print(labels.ids, labels.index, len(labels))
boxes = labels.bbox # [frame, object, x1, y1, w, h] by default
classes = labels.cls # [frame, object, 1] object arrays
visible = labels.presence # [frame, normalized-id] bool array
The text schema, count rules, ID normalization, shape semantics, DataFrame contract, and current compatibility limits are in the bundled references. Run the deterministic fixture smoke before using a real file:
python scripts/labels_fixture_smoke.py --help
python scripts/labels_fixture_smoke.py
The normal package import reaches tracking.py, which imports cv2
eagerly; install a compatible OpenCV package when import pykitti or this
module import fails with ModuleNotFoundError: cv2. The smoke script performs
only local in-memory checks and has no network or GUI behavior.
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).