← Back to Skills Marketplace
tianheihei002

Yq Video Motion Analyzer

by tianheihei002 · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ Security Clean
23
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install yq-video-motion-analyzer
Description
视频动作分析助手:从教学视频中提取关键帧、分析动作姿态、生成火柴人动作示意图。适用于运动教学视频的逐帧拆解、姿态分析和改进建议。触发关键词:视频分析、动作分析、运动私教、火柴人、姿态分析、截帧
README (SKILL.md)

视频动作分析助手

Overview

专业的视频动作分析助手,帮助用户分析教学视频中的动作姿态,并生成火柴人动作示意图。上传一段运动视频,AI 帮你逐帧拆解动作、分析姿态、指出问题——就像身边有个 24 小时在线的专业教练!

工作流程

严格按以下 3 个步骤顺序执行

步骤1:视频截帧(关键帧提取)

任务:从用户上传的视频中提取关键帧,为后续分析做准备。

工具bash 运行 Python 脚本

命令格式

python ./script/extract_frames.py \x3Cvideo_path> \x3Coutput_dir> \x3Cinterval>

参数说明

  • video_path: 视频文件的绝对路径
  • output_dir: 输出目录,默认为 output/frames
  • interval: 截帧间隔(秒),默认为 1.0

典型调用

python ./script/extract_frames.py \x3Cvideo_path> output/frames 1.0

输出

  • 提取的关键帧图片保存到 output/frames/
  • 文件命名格式:frame_XXXX_时间戳.jpg

步骤2:动作分析

任务:分析视频和关键帧中的动作姿态。

工具videos_understand + process_images(理解模式)

⚠️ 前置操作(必须):先执行 load_toolkit image_tool

操作流程

2.1 视频整体分析

使用 videos_understand 工具分析视频,获取动作概览。

分析内容:

  • 识别主要动作类型和阶段
  • 标注动作的开始/结束时间
  • 提取技术要点

2.2 关键帧姿态分析

使用 process_images 工具的 理解模式 分析关键帧。

分析内容:

  • 身体部位位置和角度(头、肩、臂、躯干、髋、腿、脚)
  • 动作质量评估(完成度、协调性、平衡性)
  • 问题识别(姿态偏差、发力错误、重心不稳)

输出格式

返回包含以下信息的分析结果:

{
  "overview": "动作概述",
  "phases": [
    {
      "name": "阶段名称",
      "time": "时间段",
      "keyPoints": ["要点1", "要点2"],
      "issues": ["问题1"]
    }
  ],
  "evaluation": {
    "score": 85,
    "strengths": ["优点1", "优点2"],
    "improvements": ["改进建议1", "改进建议2"]
  }
}

步骤3:火柴人生成

任务:根据分析结果生成火柴人骨架动作示意图,直观展示动作姿态和改进建议。

工具process_images(生成模式)

风格要求

  • 主体:火柴人骨架,线条简洁清晰
  • 背景:纯黑色背景
  • 配色
    • 火柴人:白色或高对比度彩色
    • 正确动作:绿色箭头/线条
    • 需改进:红色箭头/线条
  • 标注:关节角度、重心位置、动作方向

Prompt 模板

生成火柴人骨架动作示意图:
- 黑色背景,白色火柴人
- 绿色箭头标注正确动作方向
- 红色箭头标注需改进部分
- 显示关节角度和重心位置

动作:[动作描述]
阶段:[动作阶段]
优点:[优点,绿色标注]
改进:[改进建议,红色标注]

输出

  • 保存路径:output/stickman/
  • 格式:PNG 图片

输出文件

类型 路径
关键帧 output/frames/
火柴人图 output/stickman/

注意事项

  1. 顺序执行:严格按步骤 1 → 2 → 3 执行,不可跳过或乱序
  2. 工具加载:动作分析前必须执行 load_toolkit image_tool
  3. 截帧脚本:使用技能目录下的 ./script/extract_frames.py,该脚本依赖 opencv-pythoncv2
  4. 输出目录:确保 output/frames/output/stickman/ 目录存在或可自动创建
Usage Guidance
Install this if you intentionally want an assistant to process sports or instructional videos. Be aware that it will read the video path you provide, create extracted frame images locally, and may send video/frame content through the configured video and image understanding tools depending on the host environment.
Capability Assessment
Purpose & Capability
The stated purpose, workflow, and included Python script all align around motion analysis: read a user-supplied video, save extracted frames, analyze posture, and generate stick-figure diagrams.
Instruction Scope
The trigger terms include broad phrases like video analysis and motion analysis, so users may want to invoke it deliberately only for sports or instructional video review.
Install Mechanism
The artifact contains a markdown skill file, metadata, and one small Python helper; no package installer, startup hook, or hidden install behavior is present.
Credentials
Use of bash/Python, OpenCV, video understanding, and image processing is proportionate to frame extraction and pose analysis of uploaded media.
Persistence & Privilege
The skill writes derived image files under output/frames and output/stickman, but shows no credential access, privilege escalation, background persistence, deletion, or network exfiltration.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install yq-video-motion-analyzer
  3. After installation, invoke the skill by name or use /yq-video-motion-analyzer
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
yq-video-motion-analyzer 1.0.0 - Initial release of the video motion analysis assistant. - Supports extracting key frames from sports teaching videos for detailed action breakdown. - Provides pose analysis with feedback, strengths, and targeted improvement suggestions. - Generates stick-figure illustrations to visually represent body positions, correct actions (green), and areas needing improvement (red). - Output includes key frame images and stickman diagrams, organized in dedicated folders.
Metadata
Slug yq-video-motion-analyzer
Version 1.0.0
License MIT-0
All-time Installs 1
Active Installs 1
Total Versions 1
Frequently Asked Questions

What is Yq Video Motion Analyzer?

视频动作分析助手:从教学视频中提取关键帧、分析动作姿态、生成火柴人动作示意图。适用于运动教学视频的逐帧拆解、姿态分析和改进建议。触发关键词:视频分析、动作分析、运动私教、火柴人、姿态分析、截帧. It is an AI Agent Skill for Claude Code / OpenClaw, with 23 downloads so far.

How do I install Yq Video Motion Analyzer?

Run "/install yq-video-motion-analyzer" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Yq Video Motion Analyzer free?

Yes, Yq Video Motion Analyzer is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Yq Video Motion Analyzer support?

Yq Video Motion Analyzer is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Yq Video Motion Analyzer?

It is built and maintained by tianheihei002 (@tianheihei002); the current version is v1.0.0.

💬 Comments