Introduction

This repository contains the code for the paper "XEdgeAI: A Human-centered Industrial Inspection Framework with Data-centric Explainable Edge AI Approach" submitted to Information Fusion/Elsevier.

Requirements

Install the required libraries using the following command:

pip install -r requirements.txt

Usage

The code is written in Python and is based on the PyTorch library. The code is used to train and evaluate the proposed XAI-integrated Visual Quality Inspection Framework. The code is organized as follows:

• data/: contains the dataset used in the paper.
• model/: contains the implementation of the proposed XAI-integrated Visual Quality Inspection Framework.
  • split_dataset.py: splits the dataset into training and testing sets.
  • retrain.py: finetune the model.
  • evaluate.py: evaluate the model.
  • mobile_opt.py: optimize the model for deployment.
• explainer/: contains the implementation of the proposed XAI methods used in the paper.
  • cam.py: explain the model using the CAM methods.
  • rise.py: explain the model using the RISE method.
  • metrics.py: calculate the plausibility and faithfulness of the explanations.
• logs/: contains the logs of the training process.
• requirements.txt: contains the required libraries.

Citation

If you find this code useful, please consider citing our paper :) Thank you!

@article{nguyen2024xedgeai,
  title={XEdgeAI: A Human-centered Industrial Inspection Framework with Data-centric Explainable Edge AI Approach},
  author={Nguyen, Truong Thanh Hung and Nguyen, Phuc Truong Loc and Cao, Hung},
  journal={arXiv preprint arXiv:2407.11771},
  year={2024}
}