Metadata-Version: 2.4
Name: mvmp
Version: 0.2.10
Summary: 3D Multi-View MediaPipe - Facial landmark detection for 3D meshes
Author-email: Giuseppe Facchi <giuseppe.facchi@outlook.com>
License: MIT License
        
        Copyright (c) 2026 Giuseppe Facchi
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
        
Project-URL: Homepage, https://github.com/gfacchi-dev/mvmp
Project-URL: Repository, https://github.com/gfacchi-dev/mvmp
Project-URL: Issues, https://github.com/gfacchi-dev/mvmp/issues
Keywords: face,landmarks,3d,mediapipe,mesh
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3.11
Classifier: Topic :: Scientific/Engineering :: Image Recognition
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: mediapipe>=0.10.14
Requires-Dist: numpy<2.0,>=1.26.0
Requires-Dist: open3d>=0.18.0
Requires-Dist: opencv-python>=4.11.0.86
Requires-Dist: scikit-learn>=1.3.0
Requires-Dist: scipy>=1.11.0
Provides-Extra: dev
Requires-Dist: pytest; extra == "dev"
Requires-Dist: black; extra == "dev"
Requires-Dist: ruff; extra == "dev"
Dynamic: license-file

# MVMP: 3D Multi-View MediaPipe

[![License](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) [![Framework](https://img.shields.io/badge/Framework-Python_3.11-yellow)](https://www.python.org/downloads/release/python-3110/) [![Face Landmarker](https://img.shields.io/badge/Model-MediaPipe_Face_Landmarker-red)](https://developers.google.com/mediapipe/solutions/vision/face_landmarker)

## Description

MVMP (Multi-View MediaPipe) is a lightweight tool for 3D facial landmark detection on static textured meshes. It renders multiple camera views of the mesh, detects 2D landmarks with MediaPipe, and backprojects them into 3D space through DBSCAN-based consensus triangulation. The result is 478 facial landmarks aligned with the 3D mesh geometry, with robust outlier rejection.

**Supported mesh formats:** .obj, .ply, .stl, .gltf, .glb, .off

<!--![alt text](./img/pipelineOverview.png)-->
<img src="./img/pipelineOverview.png">

## Installation

```bash
pip install mvmp
```

The MediaPipe Face Landmarker model is bundled in the package.

### From Source

```bash
git clone https://github.com/gfacchi-dev/mvmp.git
cd mvmp
pip install .
```

## Usage

### Python API

```python
from mvmp import Facemarker

# Create a detector
marker = Facemarker()

# Detect landmarks on a mesh
result = marker.predict("path/to/mesh.obj")
print(result)  # FacemarkerResult(478 landmarks, 478 vertex indices)

# Access results
landmarks_3d = result.landmarks_3d          # list of [x, y, z] coordinates (original scale)
vertex_indices = result.closest_vertices_ids  # closest mesh vertex per landmark

# Save to JSON
result.save_json("landmarks.json")
```

#### More projections = more accuracy

```python
marker = Facemarker(projections=500)
result = marker.predict("mesh.obj")
```

#### Custom camera angles

Instead of random projections, specify exact (yaw, pitch) angles in degrees:

```python
marker = Facemarker(camera_angles=[
    (0, 0),       # front view
    (30, 0),      # 30 degrees right
    (-30, 0),     # 30 degrees left
    (0, -20),     # looking up
    (0, 15),      # looking down
])
result = marker.predict("mesh.obj")
```

#### Process multiple meshes

```python
marker = Facemarker(projections=200)

for mesh_path in mesh_files:
    result = marker.predict(mesh_path)
    result.save_json(f"output/{mesh_path.stem}.json")
```

#### Quiet mode

```python
marker = Facemarker(verbose=False)
result = marker.predict("mesh.obj")
```

### Command Line

```bash
mvmp path/to/mesh.obj -p 100 -o output/

# Process all mesh files in a directory (supports .obj, .ply, .stl, .gltf, .glb, .off)
mvmp meshes/ -p 200 -o results/
```

**Arguments:**
- `path`: Path to mesh file or directory
- `-p, --projections-number`: Number of projections (default: 500)
- `-o, --output-path`: Output directory

### Output Format

JSON output contains coordinates at the original mesh scale:

```json
{
  "coordinates": [[x, y, z], ...],
  "closest_vertex_indexes": [idx1, idx2, ...]
}
```

### Results
<!--![alt text](./img/results.png)-->
<img src="results.png">

## Contributing

1. Fork the repository and create a feature branch.
2. Make your changes with clear commit messages.
3. Open a pull request.

## License

[MIT License](LICENSE)

## Contact

Questions or suggestions? Open an issue on [GitHub](https://github.com/gfacchi-dev/mvmp/issues).
