Skip to content

Spotter

An AI coaching system that is available via app to consumers or a multi-camera system that businesses (like gyms or top golf) can install at their facility to give their members specific AI based feedback

Code Video Cerebral Valley

Core trick

MediaPipe extracts per-frame pose landmarks, rule checks compute depth and knee-tracking angles, and Gemini turns them into coaching text.

Tags

healthvision

Description

An AI coaching system that is available via app to consumers or a multi-camera system that businesses (like gyms or top golf) can install at their facility to give their members specific AI based feedback
kunal-bajpai/spotter★ 00 forksPythonCreated May 23, 2026 · last push May 27, 2026

README excerpt

# Spotter

AI squat coach. Upload a video of a back squat and get pose-estimated kinematic feedback plus a corrected-form video.

The pipeline runs as four agents:

1. **Vision** — MediaPipe extracts pose landmarks frame-by-frame.
2. **Diagnostics** — kinematic checks (depth, knee tracking, torso angle, tempo).
3. **Coach** — Gemini turns the diagnostics into plain-English feedback.
4. **Synthesis** — generates a corrected-form demo video.

Architectural detail lives in [AGENTS.md](agents.md).

## Requirements

- Python 3.12 (managed via [`uv`](https://docs.astral.sh/uv/))
- A Gemini API key — get one at https://aistudio.google.com/apikey

## Setup

```bash
uv sync                 # creates .venv, installs deps
cp .env.example .env    # then edit .env and paste your Gemini key
```

## Run

```bash
uv run streamlit run app.py
```

The app opens at <http://localhost:8501>.

## Layout

```
app.py            Streamlit UI + orchestration entrypoint
src/              Agent implementations
  vision_agent.py
  diagnostic_agent.py
  coach_agent.py
  synthesis_agent.py
  orchestrator.py
  utils/logger.py
tests/            pytest suite
data/             input videos (git-ignored)
models/           MediaPipe weights (git-ignored)
cache/            temp uploads and intermediate artifacts (git-ignored)
outputs/          generated corrected-form videos (git-ignored)
logs/             app.log (git-ignored)
```

## Tests

```bash
uv run pytest
```

Team