AGENT SKILL

detection-and-ai-components

Guides Viseron motion detection, object detection, post-processing, labels, masks, zones, and AI detector hardware or service choices.

License 许可
MIT
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
detection-and-ai-components
license
MIT
metadata
{ "disco-role": "operating" }
description
Guides Viseron motion detection, object detection, post-processing, labels, masks, zones, and AI detector hardware or service choices.
disable-model-invocation
true

02 资源文件 5

03 SKILL.md 正文

Detection and AI Components

Use this sub-skill when the user is configuring or diagnosing Viseron's AI pipeline: motion detectors, object detectors, face recognition, image classification, license plate recognition, labels, filters, masks, zones, post-processing triggers, model paths, detector devices, or external AI services.

Route by task

  • Choose a detector or backend: read detector selection before editing component blocks such as background_subtractor, mog2, mqtt, yolo, darknet, edgetpu, hailo, codeprojectai, deepstack, dlib, or compreface.
  • Explain why detections do or do not trigger recordings: read labels, masks, zones, and filters, then check the candidate config with scripts/check_detection_config.py.
  • Configure face recognition, image classification, license plate recognition, training folders, or object-triggered post-processors: read post-processing and training.
  • Diagnose optional hardware, service, model, payload, motion-overlap, mask, or label failures: read troubleshooting.

Boundaries

This sub-skill owns detector and post-processor behavior only. Route camera URL acquisition, FFmpeg/GStreamer stream construction, NVR storage tiers, recordings, snapshots, timelapse, and live relay to camera-recording-pipeline. Route Docker installation, config-file basics, secrets, logging, webserver, and safe startup to configuration-and-deployment. Route MQTT entity publication, Home Assistant discovery, webhooks, notifications, PTZ commands, and template delivery to automation-and-integrations. Route writing new components/domains/tests or changing Viseron source to developer-extension.

Safe workflow defaults

  1. Treat GPU, EdgeTPU, Hailo, external AI services, camera streams, containers, model downloads, and credentials as target-host requirements unless the user explicitly asks for live verification on that host.
  2. Validate small detector snippets with the bundled checker before restarting Viseron; it parses YAML/JSON and never imports Viseron, OpenCV, ML frameworks, or service clients.
  3. For recording-related questions, inspect the full path: object detector scanning gate, label/zone filters, trigger_event_recording, scan_on_motion_only, motion detector presence, and motion-overlap requirements.
  4. For post-processors, first ensure the object detector tracks the source object labels; post-processors only receive objects that passed object-detector filtering.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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).

★ 351