DemoSemanticSegmentationADAS

This repository provides a simple baseline Semantic Segmentation demo used in the UniPR ADAS course. It demonstrates how to train and test semantic segmentation models using the CityScapes dataset.

---

Table of Contents

• Setup
• Dataset
• Folder Structure
• Running the Demo
• Configuration

---

Setup

1. Clone the repository into your projects directory:

git clone https://github.com/MarcelloCeresini/DemoSemanticSegmentationADAS.git
cd DemoSemanticSegmentationADAS

2. Create a data directory inside the project folder:

mkdir data

3. Set up a Python virtual environment and install dependencies:

python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate
pip install -r requirements.txt

If you want GPU acceleration, install PyTorch with CUDA support following the instructions here.

---

Dataset

This project uses the CityScapes dataset.

1. Create an account with your university email.

2. Download the following splits:

   • gtFine_trainvaltest
   • leftImg8bit_trainvaltest

---

Folder Structure

After downloading and extracting the dataset, organize your files as follows:

DemoSemanticSegmentationADAS/
├── main.py
├── requirements.txt
└── data/
    ├── gtFine_trainvaltest/
    │   └── gtFine/
    │       ├── train/
    │       ├── val/
    │       └── test/
    └── leftImg8bit_trainvaltest/
        └── leftImg8bit/
            ├── train/
            ├── val/
            └── test/

---

Running the Demo

Launch the demo using:

python main.py

---

Configuration

You can modify parameters directly in main.py to experiment with different settings:

• Batch size
• Model architecture
• Number of epochs
• Learning rate

This allows you to observe how changes affect training and evaluation results.

---

Remember: Always activate the virtual environment before running the demo to ensure correct dependencies are used.