UNav
UNav is designed for helping navigation of visually impaired people. It leverages various features and algorithms to provide accurate localization and navigation assistance.
Features
- Global Feature Extraction: Using NetVLAD for image-based localization.
- Local Feature Extraction: Using SuperPoint and SuperGlue for detailed image matching.
- Navigation and Localization: Combines global and local features for accurate navigation.
Developer Installation
To install UNav as a developer, follow these steps:
Prerequisites
- Python 3.6 or higher
- Cython
- NumPy
- PyTorch
- OpenCV
- h5py
- scikit-image
Instructions
-
Clone the repository:
git clone https://github.com/ai4ce/UNav_Server.git
cd UNav_Server
-
Build the Cython extensions in place:
python setup_so.py build_ext --inplace
-
Create the source distribution and wheel:
python setup.py sdist bdist_wheel
-
Upload the package to PyPI:
twine upload --verbose dist/*
User Installation
Users can directly install UNav from PyPI:
pip install unav