Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

YOLO 视频流实时推理 · yangge521 | DecimalAI

ai mlby yangge521

YOLO 视频流实时推理

yangge521/yolo-video-inference

Source yangge521/dev-skills-zh ↗License MIT

YOLO 视频流实时推理方案，基于 Ultralytics 框架，支持 YOLOv8/YOLO11 多路并发推流

SkillScore91Maintained

Automatic

AI Engineer pack

Last verified 8/3/2026

What it changes

Passed 20 of 22 test cases with the skill, and 14 of 22 without it.

Without skill64%

With skill91%

Checks it passed that the model alone missed

States an expected performance increase of 30% to 50%.
Recommends a maximum concurrency limit of 2 video streams per GPU card.
Identifies GPU 0 and GPU 4 as the abnormal cards under multi-stream load.

and 5 more in the full report

7 fixed13 passed either way1 still failing1 broke

Pass-rate lift: +27 pp
Tokens: +49%
Agent turns: 0%
Response time: -14%

Model gemini-3.6-flashMeasured 8/3/2026v120 of 22 cases passed with skill

Tokens, turns and time describe resource use, not a dollar price.

Use this skill

Save it into your project's skills folder. Claude Code picks it up in its next session.

curl -fsSL --create-dirs -o .claude/skills/yangge521-yolo-video-inference/SKILL.md https://app.decimal.ai/s/yangge521-yolo-video-inference/SKILL.md

Want it pinned and updatable? pip install "decimalai>=0.15", then decimalai skills add yangge521-yolo-video-inference, and start Claude Code with claude --plugin-dir "$(git rev-parse --show-toplevel)/.agents"

1 · One-time — install the CLI (docs ↗)

pip install "decimalai>=0.15"

2 · Get the skill — writes .agents/skills/yangge521-yolo-video-inference/ and pins it in .decimal/skills.lock

decimalai skills add yangge521-yolo-video-inference

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/yangge521-yolo-video-inference/SKILL.md https://app.decimal.ai/s/yangge521-yolo-video-inference/SKILL.md

Raw file: /s/yangge521-yolo-video-inference/SKILL.md ↗ · machine-readable summary at /s/yangge521-yolo-video-inference.json ↗

Safety checks: 3 of 3 passed

Code scan CleanIntent PassedContent Clean

What's inside

The instructions your model reads when the skill loads.

SKILL.md120 linessingle file

YOLO 视频流实时推理

概述

基于 Ultralytics YOLOv8/YOLO11 实现视频流实时目标检测，支持 RTSP/RTMP 视频源输入，多路并发推理，RTMP 推流输出。已在 MTT S4000 (48GB) 和 NVIDIA T4 (16GB) 上完成压测验证。

使用场景

安防监控：多路摄像头实时目标检测与告警
交通分析：车流量统计、违章检测、车牌识别
工业质检：流水线产品缺陷检测
智慧工地：安全帽检测、危险区域入侵告警
零售分析：顾客行为分析、客流统计
无人值守：机房监控、设备状态检测

真实压测数据

单卡并发性能（MTT S4000, YOLO11n, 1080p@60fps）

并发路数	单路 FPS	总 FPS	稳定性
1 路	54.80	54.80	稳定
2 路	36.63	73.26	稳定
3 路	27.83	83.49	稳定
4 路	22.74	90.96	稳定

单卡最大支持 2 路 1080p@60fps 实时推理（满足 30fps 实时性要求）

多卡并发性能（8 x MTT S4000）

模式	总 FPS	达标路数	说明
8卡 x 1路	374.84	16/16	全部达标
8卡 x 2路	493.15	12/16	4路未达标

GPU 0 和 GPU 4 在 2 路并发时出现异常，建议排查散热和驱动

操作指南

1. 环境准备

Track record

Benchmark

20 / 22

test cases passed, vs 14 / 22 without

Adoption

Users: 0
Times used: 0

Safety and source details

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

View 15 scanner checks

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

yangge521 / dev-skills-zh

skills/ai-engineering/yolo-video-inference/SKILL.md

View on GitHub ↗

License MITImported 8/1/2026Version v1

More skills from the same source and nearby in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

What's inside

YOLO 视频流实时推理

概述

使用场景

真实压测数据

单卡并发性能（MTT S4000, YOLO11n, 1080p@60fps）

多卡并发性能（8 x MTT S4000）

操作指南

1. 环境准备

Track record

Benchmark

Adoption

SkillSafety

Source

Related skills