This repository provides a complete solution for semantic segmentation using a U-Net architecture trained on a subset of the Lyft Udacity Self-Driving Car Dataset (adapted as a proxy for furniture and essentials).

ml_project/
├── data/                             
│   └── dataset link.txt
├── docs/                             
│   └── architecture.txt
│   └── approach.md
├── problem_statement/
│ └── problem.md
├── solution/ 
│ └── Solution.ipynb
├── src/
│  ├── data_loader.py           
│  ├── model_unet.py             
│  ├── train_unet.py             
│  └── evaluate_unet.py          
├── tests/
│   └── test_unet.py  
├── visualisation/
│   └── visualisation.png
│   └── visualisation.png
│   └── visualisation.png 
│   └── visualisation.png
├── requirements.txt                  
└── LICENSE                           

To configure your local environment, follow these clear steps:

1. Set up a Python virtual environment

   python -m venv venv
   source venv/bin/activate  # Use venv\Scripts\activate for Windows

2. Install project dependencies

   pip install -r requirements.txt

Initiate training with the following command:

python src/train_unet.py

Ensure your dataset directories:

• RGB Images: ./data/carla/CameraRGB/
• Segmentation Masks: ./data/carla/CameraSeg/

Evaluate your trained model by importing the evaluation script:

from src.segmentation.evaluate_unet import evaluate_model

# Example usage (customize as needed)
evaluate_model(model_path='path/to/saved/model')

All libraries and dependencies required by this project are neatly organized in the requirements.txt file.

This project uses the MIT License, which allows you to:

• Freely use and modify the software commercially or non-commercially.
• Distribute copies, including modified versions.
• Integrate this software into proprietary products.

You must include the original copyright notice and a copy of the license text in any redistribution. See the LICENSE file for details.