An AI-powered automated vehicle inspection system designed for the insurance and car rental industries. Vehi-Vision leverages computer vision to detect, segment, and explain exterior vehicle damage through a transparent scoring system.
The goal of this graduation project is to create a seamless inspection pipeline that:
- Detects exterior damages (dents, scratches, broken lights, etc.) using YOLO.
- Segments specific car parts to provide localized context for the damage.
- Utilizes XAI (Grad-CAM) to generate heatmaps, explaining the model's decision-making.
- Generates a severity report and damage score based on visual findings.
Note: This project is strictly for exterior assessment; it does not cover internal mechanics or cabin interiors.
Development:
- Language: Python 3.11.x (Specifically to avoid compatibility issues)
- Detection & Segmentation: YOLO (v8/v11) and YOLO-Seg
- Explainable AI: Grad-CAM (PyTorch-based)
- Libraries: OpenCV, Pillow, NumPy, Matplotlib
Deployment
- UI Framework: Streamlit (For the functional application interface)
-
Clone the repository:
git clone https://github.com/AbdullahGhulam/Vehi-Vision cd Vehi-VisionCreate a Virtual Environment:
# This creates a new environment (run it only ONCE) python -m venv venvAfter that, activate the environment using:
source venv/bin/activate # Linux/macOS # venv\Scripts\activate # Windows
-
Install Dependencies:
pip install --upgrade pip pip install -r requirements.txt
-
Verify Installation:
After installing the requirements, run the following script to ensure your environment (Python, PyTorch, and GPU) is configured correctly:python check_setup.py
📂 Project Structure
Vehi-Vision
├── data/ # Local datasets (Ignored by Git)
├── venv/ # Python Environment (Ignored by Git)
├── weights/ # Saved model weights (.pt or .onnx)
├── notebooks # Research, EDA, and prototyping
├── src/ # Source code (Detection, Segmentation, XAI)
│ └── utils
├── .gitignore # Files to exclude from Git
└── README.md
👥 Team & Credits