Metadata-Version: 2.4
Name: elevate3d
Version: 0.4.1
Summary: 3D terrain and structure reconstruction from single RGB images
Home-page: https://github.com/krdgomer/elevate3d
Author: Ömer Can Karadağ
Author-email: krdg.omercan@hotmail.com
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: open3d
Requires-Dist: numpy
Requires-Dist: pillow
Requires-Dist: opencv-python
Requires-Dist: torch
Requires-Dist: albumentations
Requires-Dist: deepforest
Requires-Dist: scikit-image
Requires-Dist: huggingface_hub
Requires-Dist: flask
Requires-Dist: trimesh
Requires-Dist: matplotlib
Requires-Dist: scipy
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: license-file
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# Elevate3D

**Generate 3D models from satellite & aerial images using deep learning.**

---

##  Disclaimer

**Elevate3D is an experimental, early-stage project.**  
Itâ€™s not production-ready, and results may be inconsistent depending on the input. Expect rough outputs, strange artifacts, and occasional surprises.


---

##  Features

-  Automatic Building Segmentation (Mask R-CNN)  
-  Elevation Prediction from RGB (Pix2Pix)  
-  3D Mesh Generation (Open3D)  
-  Tree Detection with DeepForest  
-  Pretrained Models â€“ No training required  
-  End-to-End Pipeline â€“ From image to interactive 3D output

---

##  Installation

Install with pip:

```bash
pip install elevate3d
```

---

##  Usage

### 1. Web Interface (Recommended)

Run the local web app:

```bash
elevate3d-run
```

This launches a local server where you can upload images and view results interactively in your browser.

### 2. Python API

You can also run the pipeline programmatically:

```python
from elevate3d.run_pipeline import run_pipeline

run_pipeline("path_to_your_image.jpg")
```

This processes the image and opens a viewer window showing the 3D model.

---

##  Input Requirements

- Accepts **aerial or satellite RGB images** (`.jpg`, `.jpeg`, `.png`)
- Images should ideally be top-down and contain visible buildings or tree cover

---

##  How It Works

- **Mask R-CNN** segments buildings
- **Pix2Pix** generates DSM from RGB
- **Tree detection** adds tree geometry
- **Open3D** constructs the final mesh (as `.glb`)

---

##  License

MIT License. See [`LICENSE`](LICENSE) for details.
