pose-platform 1.25.3

Pose CLI — motion analysis toolkit. Install sports via extras: pose-platform[badminton]

Pose CLI

Open motion-analysis toolkit for sports scientists. pip install pose-platform — analyze 3D skeleton data on your machine.

pip install pose-platform
pose analyze data/session.csv --tool stats

---

What is Pose CLI?

Pose CLI turns 3D body-skeleton data into reproducible sports analytics. It runs entirely on your machine — your data never leaves your computer.

Pose Coach Recording → JointData3D CSV → pose analyze --tool <name> → JSON Output

11 built-in analysis tools, two supported sports:

Category	Tools	Sport
General (any movement)	stats, detect_peak_energy, segment_motion_phases, detect_velocity, detect_action_boundaries, detect_highlight	freeplay
Running (reference implementation)	measure_running_speed, score_sprint, analyze_running_gait, score_running_efficiency	sprint
Multi-sport	analyze_motion_angles	both

Sprint tools serve as a reference for building your own sport-specific tools. See contributing guide →

Quickstart

pip install pose-platform

# 获取示例数据（内置在包中，无需网络）
pose sample

# Run your first analysis
pose analyze sample_sprint.csv --tool stats

# Explore available tools
pose tools && pose sports

# LLM-powered analysis (optional — needs openai SDK + API key)
pip install openai>=1.50
export POSE_DEEPSEEK_API_KEY=sk-xxx
pose chat ./data

Installation

• Python 3.10+ required. Works on macOS, Linux, Windows.
• pose chat LLM 对话：需额外安装 pip install openai>=1.50，并设置 API Key：

  # DeepSeek（推荐，邮箱注册即得）
  export POSE_DEEPSEEK_API_KEY=sk-xxx
  # 或 Qwen / 其他 CN provider
  export POSE_CN_API_KEY=sk-xxx
  

  详见 pose chat 使用指南。
• Full guide: posecap.com/posecli/docs/

Data Format

Pose CLI reads JointData3D CSV (12-column, no header):

timestamp, joint, wx, wy, wz, mx, my, mz, ldq_x, ldq_y, ldq_z, ldq_w

Record with Pose Coach (iOS, TestFlight) or convert from OpenPose / MediaPipe / motion capture. Data format reference →

Commands

pose analyze <file.csv> --tool <name> [--param key=value]  # Run analysis
pose sample                                                    # Copy sample data to CWD
pose tools [--sport <id>] [--detail]                         # List tools
pose sports                                                   # List sports
pose gaps list|submit|show|summary                            # Track feature requests
pose chat [data_dir]                                          # LLM-powered analysis
pose cache list|clear                                         # Manage cache
pose telemetry stats [--since 7d]                             # Usage statistics
pose version [--changelog]                                    # Version info

Extending Pose CLI

Every analysis tool is a Python class. Build your own:

from capabilities.tools.base import DirectTool
from capabilities.tools.loader import load_pose
import json
from pathlib import Path

class MyTool(DirectTool):
    name = "my_tool"
    supported_sports = ["basketball"]  # or [] for all
    input_schema = {"type": "object", "properties": {"path": {"type": "string"}}, "required": ["path"]}

    def run(self, tool_input: dict, output_dir: Path) -> str:
        pose = load_pose(tool_input["path"])
        # your analysis here
        return json.dumps({"result": 42}, ensure_ascii=False)

Register a new sport in 3 lines of config. Contributing →

From Source (Platform Developers)

Pose CLI is the public distribution of the Pose Platform. The full platform (in this repository) also includes the Pose Coach backend, operations tooling, and badminton-specific analysis tools — these are excluded from the pip package.

git clone <repo-url>
cd pose_platform
pip install -e .

License

Pose CLI is distributed under the Pose Platform License v1.0:

• Research use: free for academic and educational purposes
• Commercial use: requires a separate agreement
• Attribution: cite the Pose Platform in publications

Citation

@software{pose_platform,
  title = {Pose Platform: Open Motion Analysis Toolkit},
  author = {{PoseCap Inc.}},
  year = {2026},
  url = {https://github.com/posecap/pose-platform},
}