Skip to content

Fratres-X-AI/SYNAPSE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

22 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Synapse logo

Synapse

Synapse is a webcam-first, local-first focus and fatigue support tool. It uses the local webcam to extract face landmarks on-device, estimates coarse attention signals in real time, and saves session summaries for review.

Synapse is pilot software. It is not a medical device, lie detector, emotion oracle, productivity scoring system, or employee discipline tool.

What It Does

  • Runs webcam processing locally with MediaPipe and OpenCV.
  • Guides each user through onboarding: attention calibration plus an optional expression profile.
  • Monitors live sessions for engagement, fatigue, tension, positivity, distraction, and broad attention state.
  • Saves local CSV logs, alert logs, and text reports for review.
  • Replays saved monitor sessions for debriefing and quality checks.

Raw webcam video is not saved by Synapse. See docs/privacy.md before any pilot use.

Setup

Use Python 3.11 or 3.12 on Windows with a working webcam. Do not use the latest Python (3.13+) or older releases such as 3.7 — mediapipe==0.10.21 only ships wheels for 3.11–3.12.

Quick start (share with pilots)

From the Synapse folder in PowerShell, run once:

.\scripts\setup_windows.ps1

That installs dependencies and creates a Synapse icon on the desktop. Double-click it to open the launcher window. Choose First Run the first time.

No PowerShell? Double-click Launch Synapse.bat after setup. If dependencies are missing, the batch file will tell you to run setup_windows.ps1.

Manual install

Check your environment before installing:

.\scripts\check_python.ps1

Recommended install:

py -3.11 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
pip install -r requirements.txt

Optional tray and packaging dependencies:

pip install -r requirements-dev.txt

If you only want to run Synapse (no Python setup), use Synapse.exe from a GitHub Release — see docs/windows_install.md.

Install troubleshooting

Symptom Likely cause Fix
No matching distribution found for mediapipe Python 3.13+ (or very old Python) Install Python 3.11, recreate .venv, rerun check_python.ps1
Install hangs or fails on mediapipe after downgrading to 3.7 Wrong direction — 3.7 is unsupported Use 3.11 or 3.12 instead
ImportError / DLL load failed on Windows Missing MSVC runtime Install Visual C++ Redistributable (x64)
numpy version conflict NumPy 2.x pulled in by another package Stay in a fresh venv; requirements.txt pins numpy<2

More detail: docs/known_issues.md.

Launcher Commands

Run commands from the repository root:

python synapse_launcher.py onboard
python synapse_launcher.py monitor
python synapse_launcher.py monitor --fullscreen
python synapse_launcher.py showcase
python synapse_launcher.py showcase --fullscreen
python synapse_launcher.py replay "%LOCALAPPDATA%\Synapse\sessions\monitor_YYYYMMDD_HHMMSS.csv"
python synapse_launcher.py --tray

onboard records privacy consent, then captures calibration values and an expression profile. monitor starts a live webcam session and writes logs/reports. showcase runs the elite landmark shell and flight HUD for demos (no session logging). replay opens a saved session CSV; pass the CSV path explicitly for app-data sessions. Press q to quit a camera window and f to toggle fullscreen where supported.

Local Data

On Windows, Synapse stores user data under:

%LOCALAPPDATA%\Synapse

Expected contents:

  • config\privacy_consent.json
  • config\calibration.json
  • config\emotion_profile.json
  • config\settings.json
  • sessions\monitor_*.csv
  • sessions\monitor_*.alerts.csv
  • sessions\monitor_*.report.txt

By default, monitor reports may also be copied to the Desktop as Synapse_Report_*.txt.

Pilot Workflow

Use the no-payment pilot process in docs/pilot_guide.md until the software quality gate is met.

  1. Explain consent and limitations.
  2. Run python synapse_launcher.py onboard.
  3. Run a short camera/environment check.
  4. Run python synapse_launcher.py monitor.
  5. Stop with q, review the generated report, and replay the session if needed.

Build

For a smoke-test Windows executable, install dev dependencies and run the checked-in build script:

pip install -r requirements-dev.txt
.\build.ps1
.\scripts\verify_release.ps1

Pilot install instructions for non-developers are in docs/windows_install.md.

Verify the build with docs/release_checklist.md before sharing it with pilot users.

Documentation

  • docs/privacy.md - privacy model, saved data, deletion/export controls, consent language.
  • docs/pilot_guide.md - 5-person no-payment pilot workflow.
  • docs/release_checklist.md - Windows release verification checklist.
  • docs/known_issues.md - current limitations and webcam constraints.
  • docs/positioning.md - approved positioning and prohibited claims.
  • docs/monetization_gate.md - no-billing rule until the quality gate is proven.

About

Synapse

Resources

Stars

1 star

Watchers

0 watching

Forks

Packages

 
 
 

Contributors