geoai-py 0.23.0__tar.gz → 0.24.0__tar.gz

{geoai_py-0.23.0 → geoai_py-0.24.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: geoai-py
-Version: 0.23.0
+Version: 0.24.0
 Summary: A Python package for using Artificial Intelligence (AI) with geospatial data
 Author-email: Qiusheng Wu <giswqs@gmail.com>
 License: MIT License

{geoai_py-0.23.0 → geoai_py-0.24.0}/geoai/__init__.py

@@ -2,7 +2,7 @@
 
 __author__ = """Qiusheng Wu"""
 __email__ = "giswqs@gmail.com"
-__version__ = "0.23.0"
+__version__ = "0.24.0"
 
 
 import os

{geoai_py-0.23.0 → geoai_py-0.24.0}/geoai_py.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: geoai-py
-Version: 0.23.0
+Version: 0.24.0
 Summary: A Python package for using Artificial Intelligence (AI) with geospatial data
 Author-email: Qiusheng Wu <giswqs@gmail.com>
 License: MIT License

{geoai_py-0.23.0 → geoai_py-0.24.0}/geoai_py.egg-info/SOURCES.txt

@@ -53,6 +53,7 @@ geoai_py.egg-info/top_level.txt
 qgis_plugin/README.md
 qgis_plugin/install.py
 qgis_plugin/install.sh
+qgis_plugin/package_plugin.py
 qgis_plugin/geoai_plugin/LICENSE
 qgis_plugin/geoai_plugin/__init__.py
 qgis_plugin/geoai_plugin/_geoai_lib.py

{geoai_py-0.23.0 → geoai_py-0.24.0}/mkdocs.yml

@@ -136,6 +136,7 @@ nav:
           - examples/water_dynamics.ipynb
           - examples/wetland_mapping.ipynb
           - examples/wetland_dynamics.ipynb
+          - examples/wetland_sam3.ipynb
           - examples/regularization.ipynb
           - examples/globe_projection.ipynb
           - examples/samgeo.ipynb

{geoai_py-0.23.0 → geoai_py-0.24.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "geoai-py"
-version = "0.23.0"
+version = "0.24.0"
 dynamic = [
     "dependencies",
 ]
@@ -45,7 +45,7 @@ universal = true
 
 
 [tool.bumpversion]
-current_version = "0.23.0"
+current_version = "0.24.0"
 commit = true
 tag = true
 

{geoai_py-0.23.0 → geoai_py-0.24.0}/qgis_plugin/README.md

@@ -4,21 +4,32 @@ A QGIS plugin that brings the [geoai](https://github.com/opengeos/geoai) models
 
 ## Quick Start
 
--   Create a fresh conda env (`conda create -n geo python=3.12`) and install QGIS + deps (see below).
--   Install the plugin (`python install.py`) from this repo.
--   Restart QGIS → `Plugins` → `Manage and Install Plugins...` → enable `GeoAI`.
+-   Create a Pixi project and install the dependencies.
+-   Install the QGIS plugin from the QGIS Plugin Manager.
+-   Enable the GeoAI plugin in QGIS.
+-   Restart QGIS.
 -   Open a GeoAI toolbar panel and try the sample datasets below.
 
 ## Video Tutorials
 
+### Installation Tutorial
+
+You can follow this [video tutorial](https://youtu.be/TJmZQXJK-IU) to install the GeoAI QGIS Plugin on Linux/Windows:
+
+[![installation](https://github.com/user-attachments/assets/6fa17c97-2ff5-40d6-be16-c6dcfe03f806)](https://youtu.be/TJmZQXJK-IU)
+
+### Usage Tutorial
+
 Check out this [short video demo](https://youtu.be/Esr_e6_P1is) and [full video tutorial](https://youtu.be/8-OhlqeoyiY) on how to use the GeoAI plugin in QGIS.
 
 [![demo](https://github.com/user-attachments/assets/5aabc3d3-efd1-4011-ab31-2b3f11aab3ed)](https://youtu.be/8-OhlqeoyiY)
 
+
+
 ## Requirements
 
 -   QGIS 3.28 or later
--   Python 3.10+ (conda recommended)
+-   Python 3.10+ (Pixi recommended)
 -   PyTorch (CUDA if you want GPU acceleration)
 -   `geoai` and `samgeo` packages
 
@@ -53,251 +64,190 @@ Each tool lives inside a dockable panel that can be attached to either side of t
 
 ## Installation
 
-### 1) Set up the environment
-
-#### Installation on Linux/macOS
+### 1. Set up the environment
 
-Use a clean conda env dedicated to QGIS—mixing with an existing QGIS install often breaks dependencies.
+Installing the GeoAI QGIS plugin on can be challenging due to the complicated pytorch/cuda dependencies. Conda or mamba might take a while to resolve the dependencies, while pip might fail to install the dependencies properly. It is recommended to use [pixi](https://pixi.prefix.dev/latest) to install the dependencies to avoid these issues.
 
-```bash
-conda create -n geo python=3.12
-conda activate geo
-```
+#### 1) Install Pixi
 
-Install core geospatial deps first:
+#### Linux/macOS (bash/zsh)
 
 ```bash
-conda install -c conda-forge --strict-channel-priority gdal rasterio libnetcdf netcdf4
-python -c "import rasterio; print('rasterio import successful')"
+curl -fsSL https://pixi.sh/install.sh | sh
 ```
 
-Install GeoAI:
+Close and re-open your terminal (or reload your shell) so `pixi` is on your `PATH`. Then confirm:
 
 ```bash
-conda install -c conda-forge geoai
-python -c "import geoai; print('geoai import successful')"
+pixi --version
 ```
 
-Install QGIS:
+#### Windows (PowerShell)
 
-```bash
-conda install -c conda-forge qgis
-```
-
-Install SamGeo extras (PyPI is required for some parts):
+Open **PowerShell** (preferably as a normal user, Admin not required), then run:
 
-```bash
-pip install -U "segment-geospatial[samgeo3]" sam3
-python -c "import samgeo; print('samgeo import successful')"
+```powershell
+powershell -ExecutionPolicy Bypass -c "irm -useb https://pixi.sh/install.ps1 | iex"
 ```
 
-#### Installation on Windows
-
-Windows installation requires some additional steps compared to Linux/macOS. Choose the appropriate section based on whether you have an NVIDIA GPU or want CPU-only installation.
-
-**Prerequisites (Required for all Windows users):**
-
-1. Install [Miniconda](https://docs.conda.io/en/latest/miniconda.html) or [Anaconda](https://www.anaconda.com/download) if you haven't already.
-2. Open **Anaconda Prompt** (not PowerShell or CMD) for all installation commands.
-3. For GPU users: Ensure you have the latest [NVIDIA GPU drivers](https://www.nvidia.com/Download/index.aspx) installed.
-
-##### Option A: Windows with NVIDIA GPU (CUDA)
-
-This option provides the best performance using your NVIDIA GPU for model inference and training.
+Close and re-open PowerShell, then confirm:
 
-**Step 1: Create and activate the conda environment**
-
-```bash
-conda create -n geo python=3.12 -y
-conda activate geo
+```powershell
+pixi --version
 ```
 
-**Step 2: Install PyTorch with CUDA support**
-
-First, check your NVIDIA driver version to determine the compatible CUDA version:
+---
 
-```bash
-nvidia-smi
-```
+#### 2) Create a Pixi project
 
-Look for the "CUDA Version" in the output. Then install the appropriate PyTorch version:
+Navigate to a directory where you want to create the project and run:
 
-For CUDA 12.4 (recommended for newer drivers):
-
-```bash
-conda install pytorch torchvision pytorch-cuda=12.4 -c pytorch -c nvidia -y
+```powershell
+pixi init geo
+cd geo
 ```
 
-For CUDA 12.1 (for older drivers):
-
-```bash
-conda install pytorch torchvision pytorch-cuda=12.1 -c pytorch -c nvidia -y
-```
+---
 
-**Step 3: Verify PyTorch GPU installation**
+#### 3) Configure `pixi.toml`
 
-```bash
-python -c "import torch; print(f'PyTorch: {torch.__version__}'); print(f'CUDA available: {torch.cuda.is_available()}'); print(f'GPU: {torch.cuda.get_device_name(0) if torch.cuda.is_available() else \"None\"}')"
-```
+Open `pixi.toml` in the `geo` directory and replace its contents with the following depending on your system.
 
-You should see `CUDA available: True` and your GPU name. If not, see the troubleshooting section below.
+If you have a NVIDIA GPU with CUDA, run `nvidia-smi` to check the CUDA version.
 
-**Step 4: Install QGIS and core dependencies**
+- For GPU with CUDA 12.x:
 
-```bash
-conda install -c conda-forge qgis -y
-```
+```toml
+[workspace]
+channels = ["https://prefix.dev/conda-forge"]
+name = "geo"
+platforms = ["linux-64", "win-64"]
 
-**Step 5: Install GeoAI**
+[system-requirements]
+cuda = "12.0"
 
-```bash
-conda install -c conda-forge geoai -y
-python -c "import geoai; print('geoai import successful')"
+[dependencies]
+python = "3.12.*"
+pytorch-gpu = ">=2.9.1,<3"
+qgis = ">=3.44.5"
+geoai = ">=0.23.0"
+segment-geospatial = ">=0.10.8"
+sam3 = ">=0.1.0.20251211"
+libopenblas = ">=0.3.30"
 ```
 
-**Step 6: Install SamGeo with SAM3 support**
-
-```bash
-pip install -U triton-windows
-pip install -U "segment-geospatial[samgeo3]"
-pip install -U sam3
-python -c "import samgeo; print('samgeo import successful')"
+- For GPU with CUDA 13.x:
+
+```toml
+[workspace]
+channels = ["https://prefix.dev/conda-forge"]
+name = "geo"
+platforms = ["linux-64", "win-64"]
+
+[system-requirements]
+cuda = "13.0"
+
+[dependencies]
+python = "3.12.*"
+pytorch-gpu = ">=2.9.1,<3"
+qgis = ">=3.44.5"
+geoai = ">=0.23.0"
+segment-geospatial = ">=0.10.8"
+sam3 = ">=0.1.0.20251211"
+libopenblas = ">=0.3.30"
 ```
 
-##### Option B: Windows CPU-Only (No GPU)
-
-Use this option if you don't have an NVIDIA GPU or want a simpler installation.
-
-**Step 1: Create and activate the conda environment**
-
-```bash
-conda create -n geo python=3.12 -y
-conda activate geo
+- For CPU:
+
+```toml
+[workspace]
+channels = ["https://prefix.dev/conda-forge"]
+name = "geo"
+platforms = ["linux-64", "win-64"]
+
+[dependencies]
+python = "3.12.*"
+pytorch-cpu = ">=2.9.1,<3"
+qgis = ">=3.44.5"
+geoai = ">=0.23.0"
+segment-geospatial = ">=0.10.8"
+sam3 = ">=0.1.0.20251211"
+libopenblas = ">=0.3.30"
 ```
 
-**Step 2: Install PyTorch (CPU version)**
+---
 
-```bash
-conda install pytorch torchvision cpuonly -c pytorch -y
-```
+#### 4) Install the environment
 
-**Step 3: Verify PyTorch installation**
+From the `geo` folder:
 
-```bash
-python -c "import torch; print(f'PyTorch: {torch.__version__}'); print('PyTorch CPU installation successful')"
+```powershell
+pixi install
 ```
 
-**Step 4: Install QGIS and core dependencies**
-
-```bash
-conda install -c conda-forge qgis -y
-```
+This step may take several minutes on first install depending on your internet connection and system.
 
-**Step 5: Install GeoAI**
+---
 
-```bash
-conda install -c conda-forge geoai -y
-python -c "import geoai; print('geoai import successful')"
-```
+#### 5) Upgrade `segment-geospatial` via pip
 
-**Step 6: Install SamGeo (without SAM3)**
+Install the latest `segment-geospatial` release on top of the resolved environment:
 
-```bash
-pip install segment-geospatial
-python -c "import samgeo; print('samgeo import successful')"
+```powershell
+pixi run pip install -U segment-geospatial
 ```
 
-##### Windows Troubleshooting
+---
 
-**Common Issue 1: CUDA not detected after PyTorch installation**
+#### 6) Verify PyTorch + CUDA
 
-If `torch.cuda.is_available()` returns `False`:
+If you have a NVIDIA GPU with CUDA, run the following command to verify the PyTorch + CUDA installation:
 
-1. Verify NVIDIA drivers are installed: Run `nvidia-smi` in command prompt
-2. Ensure you installed the CUDA-enabled PyTorch (not CPU version)
-3. Try reinstalling PyTorch:
-
-```bash
-conda uninstall pytorch torchvision -y
-conda install pytorch torchvision pytorch-cuda=12.4 -c pytorch -c nvidia -y
+```powershell
+pixi run python -c "import torch; print('PyTorch:', torch.__version__); print('CUDA available:', torch.cuda.is_available()); print('GPU:', (torch.cuda.get_device_name(0) if torch.cuda.is_available() else 'None'))"
 ```
 
-**Common Issue 2: DLL load failed or missing dependencies**
+Expected output should be like this:
 
-If you see errors like `DLL load failed` or `ImportError`:
+-   `PyTorch: 2.9.1`
+-   `CUDA available: True`
+-   `GPU: NVIDIA RTX 4090`
 
-1. Install Microsoft Visual C++ Redistributable:
-    - Download and install [VC++ Redistributable](https://aka.ms/vs/17/release/vc_redist.x64.exe)
-2. Restart your computer after installation
+If CUDA is `False`, check:
 
-**Common Issue 3: Triton installation fails**
+-   `nvidia-smi` works in PowerShell
+-   NVIDIA driver is up to date
 
-Triton is required for SAM3 on Windows. If `pip install triton-windows` fails:
-
-1. Ensure you're using Python 3.12 (not 3.13+)
-2. Try installing from conda-forge:
-
-```bash
-pip install triton-windows --no-cache-dir
-```
+---
 
-If Triton still doesn't work, you can skip SAM3 and use SAM1/SAM2 instead.
 
-**Common Issue 4: Permission errors during installation**
+#### Request access to SAM 3
 
-Run Anaconda Prompt as Administrator, or try:
+To use SAM 3, you will need to request access by filling out this form on Hugging Face at <https://huggingface.co/facebook/sam3>. Once your request has been approved, run the following command in the terminal to authenticate:
 
 ```bash
-pip install --user <package-name>
+pixi run hf auth login
 ```
 
-**Common Issue 5: QGIS fails to start or shows import errors**
-
-Make sure you launch QGIS from the activated conda environment:
+After authentication, you can download the SAM 3 model from Hugging Face:
 
 ```bash
-conda activate geo
-qgis
+pixi run hf download facebook/sam3
 ```
 
-Do NOT use the QGIS shortcut from the Start Menu—it won't have access to the conda packages.
-
-**Common Issue 6: Out of memory errors**
-
-If you run out of GPU memory:
+**Important Note**: SAM 3 currently requires a NVIDIA GPU with CUDA support. You won't be able to use SAM 3 if you have a CPU only system ([source](https://github.com/facebookresearch/sam3/issues/164)). You will get an error message like this: `Failed to load model: Torch not compiled with CUDA enabled`.
 
-1. Use the **GPU** button in the GeoAI toolbar to clear memory
-2. Close other GPU-intensive applications
-3. Use smaller batch sizes in training/inference settings
-4. Switch to CPU mode in the plugin settings for smaller tasks
-
-##### Video Tutorial
-
-You can follow this [video tutorial](https://youtu.be/a-Ns9peiuu8) to install the GeoAI QGIS Plugin on Windows:
-
-[![windows](https://github.com/user-attachments/assets/8d89d535-1d66-45d2-a6c0-171416c259c9)](https://youtu.be/a-Ns9peiuu8)
-
-#### Request access to SAM 3
-
-To use SAM 3, you will need to request access by filling out this form on Hugging Face at <https://huggingface.co/facebook/sam3>. Once your request has been approved, run the following command in the terminal to authenticate:
-
-```bash
-hf auth login
-```
-
-### 2) Install the QGIS plugin
+### 2. Install the QGIS plugin
 
 Option A — use QGIS Plugin Manager (recommended):
 
-GeoAI is available as an experimental plugin in the official [QGIS plugin repository](https://plugins.qgis.org/plugins/geoai). To install:
-
-1. Launch QGIS: `conda run qgis`
-2. Go to `Plugins` → `Manage and Install Plugins...` → `Settings` tab → check `Show also Experimental Plugins` → Click on `Reload all Repositories` button.
-3. Switch to the `All` tab, search for `GeoAI`, select it, and click `Install Experimental Plugin`.
+GeoAI is available as a QGIS plugin in the official [QGIS plugin repository](https://plugins.qgis.org/plugins/geoai). To install:
 
-![](https://github.com/user-attachments/assets/c15c4b7e-1e8c-45de-9127-671c7c2c85a9)
+1. Launch QGIS: `pixi run qgis`
+2. Go to `Plugins` → `Manage and Install Plugins...`
+3. Switch to the `All` tab, search for `GeoAI`, select it, and click `Install Plugin`
 
-![](https://github.com/user-attachments/assets/ea101a82-5df7-4947-99b8-d4a83a1598ed)
+![](https://github.com/user-attachments/assets/b31d1d13-27ff-420a-84ab-9cc82ade9a8e)
 
 Option B — use the helper script:
 
@@ -320,9 +270,9 @@ Option C — manual copy:
     -   Windows: `C:\Users\<username>\AppData\Roaming\QGIS\QGIS3\profiles\default\python\plugins\`
     -   macOS: `~/Library/Application Support/QGIS/QGIS3/profiles/default/python/plugins/`
 
-### 3) Enable in QGIS
+### 3. Enable in QGIS
 
-Launch QGIS: `conda run qgis`
+Launch QGIS: `pixi run qgis`
 
 QGIS → `Plugins` → `Manage and Install Plugins...` → enable `GeoAI`. After updates, toggle the plugin off/on or restart QGIS to reload.
 
@@ -359,7 +309,7 @@ Sample datasets:
 
 Steps:
 
-1. Download the sample datasets (links above) or prepare your own imagery/vector labels. Store them in a folder that is accessible to the conda environment.
+1. Download the sample datasets (links above) or prepare your own imagery/vector labels. Store them in a folder that is accessible to pixi project.
 2. Click the **Segmentation** button in the GeoAI toolbar (or `GeoAI` menu → `Segmentation`)
 3. Use the tabs at the top of the panel to switch between:
 
@@ -467,7 +417,6 @@ The QGIS plugin supports any models supported by [Pytorch Segmentation Models](h
 ## Troubleshooting
 
 -   Plugin missing after install: confirm the plugin folder exists in your QGIS profile path and that you restarted QGIS.
--   GDAL/rasterio errors: verify you launched QGIS from the conda env (`conda activate geo` then `qgis`) so it picks up the same Python libs.
 -   CUDA OOM: use the **GPU** button to clear cache, lower batch sizes, or switch to CPU for smaller runs.
 -   Model download failures: check network/firewall, then retry loading models from the panel.