Instructions

Code for Assistive Signals in DNN Classifiers:

1. We can install the Python libraries needed by running in the command line:

   pip install -r requirements.txt

2. Additionally, we used Pytorch3D v.0.4.0 and Pytorch. Follow the corresponding instruction to install the latest Pytorch3D version: https://github.com/facebookresearch/pytorch3d

3. You can run the code to generate the FUll Assistive Textures by running in the command line:

   sudo python run_optimization.py

   Optionally: You can use Jupyter Notebooks to visualize the models created in run_optimization.ipynb or check directly in the outputs folder.