ChronoFace

Windows desktop application that sorts a folder of photos by the estimated age of a specific person appearing in them.

Sort large, unordered photo collections by a specific person's estimated age — youngest to oldest — for presentations, slideshows, albums, and other chronological workflows.

Privacy

All photo analysis is performed locally on this computer. No photos or facial data are uploaded.

Status

MVP complete through review + export, plus Settings → model packs:

• Local analysis (metadata, faces, age, ranking)
• Review & Correct timeline (drag order, manual age, face reassignment, exclude)
• Export to Folder numbered age-ordered copies + CSV
• Settings to choose OpenCV Fast or InsightFace buffalo/antelope packs
• Windows portable build script (scripts/build_windows.py)

Requirements

• Windows 10/11 (macOS later)
• Python 3.11–3.13 (recommended: 3.11; see .python-version)

If Python is missing (or C:\Python313 is incomplete / missing Lib\):

choco install python311 -y

Open a new terminal so py -3.11 is on PATH, then run setup.bat.

Setup (Windows)

One-shot (creates/repairs .venv and installs deps with matching wheels):

git clone https://github.com/OptimaLPro/ChronoFace.git
cd ChronoFace
setup.bat

Manual:

py -3.11 -m venv .venv
.venv\Scripts\activate
python -m pip install -r requirements.txt

Git Bash:

git clone https://github.com/OptimaLPro/ChronoFace.git
cd ChronoFace
./setup.sh

setup.bat / setup.sh detect a broken venv (e.g. NumPy built for 3.11 but running under 3.13) and recreate it. Use setup.bat --force to always rebuild.

InsightFace packs (buffalo_l recommended) are for personal / non-commercial use. Open Settings → Models to compare speed vs quality, then Downloads to fetch packs.

Run

run.bat

Or with the venv activated:

.venv\Scripts\activate
python app.py

Do not use source in Command Prompt — that is a bash command. In CMD use .venv\Scripts\activate.

Models (Settings)

Pack	Speed	Quality	Notes
OpenCV Fast (YuNet + SFace)	Fastest	Good	Easiest; tiny download
InsightFace buffalo_s	Fast	Better	Balanced
InsightFace buffalo_l ★	Medium	Best	Recommended personal default
InsightFace antelopev2	Slow	Maximum	Largest / strongest

After changing models, run Analyze Photos again (embeddings are not compatible across packs).

Review & export

1. Click Review & Correct
2. Drag thumbnails to fix order (filter must be “All photos” to save full order)
3. Select a photo → set manual age / reassign face / approve / exclude / mark not target
4. Click Export to Folder for numbered copies in youngest-to-oldest order

Windows portable build

setup.bat
.venv\Scripts\python.exe scripts\build_windows.py

Then run dist\ChronoFace\ChronoFace.exe.

Development phases

Phase	Focus
1	Foundation, UI setup, SQLite
2	Metadata pipeline, hashing, thumbnails
3	Face detection / recognition
4	Age estimation
5	Ranking / grouping
6	Manual review timeline
7	Numbered export + CSV
8	Windows packaging script
9	Settings + swappable model packs (current)

Architecture notes

• UI (PySide6) stays separate from vision, metadata, sorting, and export logic.
• Vision models sit behind abstract interfaces so weights can be swapped when licensing allows.
• Original photos are never modified; exports are copies into the chosen output folder.

License note

Before bundling any pretrained face/age model, verify code license, model-weight license, commercial use, and redistribution rights separately.