README

Prerequisites

• Python 3.11 - 3.12
• CUDA-capable GPU with at least 4GB VRAM (recommended)
• Webcam (recommended)

Installation

1. Clone the repository:

git clone https://gitlab.lrz.de/simon-hampp/adl-gruppe-1.git
cd adl-gruppe-1

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate

3. Install dependencies:

• If using a CUDA-Compatible GPU (optional, but running a Stable Diffusion on CPU may take up to 1h):

pip install -r requirements_cuda.txt

• General:

pip install -r requirements.txt

4. Make sure Pandoc, GTK3 and MikTeX (xelatex) are installed and added to the PATH.

It should happen automatically if you are using Linux.

Otherwise you can download it from the following link:

Pandoc: https://pandoc.org/installing.html GTK3: Windows: github.com/tschoonj/GTK-for-Windows-Runtime-Environment-Installer MacOS: https://www.gtk.org/docs/installations/macos/ MiKTeX: https://miktex.org/download

Usage

1. Navigate to project root:

cd path/to/adl-gruppe-1

2. Run the pipeline:

python Code/pipeline.py [--mode <camera|images>] [--images_path /path/to/images]

Arguments:

• --mode: Specifies the mode to run the pipeline. It can be either camera or images.

  • camera: Uses the webcam to capture images. This is the default mode and can be omitted if you want to use the webcam.

  • images: Uses pre-existing images from a specified directory.

• --images_path: Specifies the path to the directory containing images to be used instead of capturing new ones with the webcam. This argument is optional and defaults to Code/webcam/demo_images, which contains images of a BMW X5. Use this argument if you want to specify a different directory.

Examples:

• To run the pipeline using the webcam (default mode):

python Code/pipeline.py

• To run the pipeline using images from the default demo directory:

python Code/pipeline.py --mode images

• To run the pipeline using images from a specific directory:

python Code/pipeline.py --mode images --images_path /path/to/images

3. Follow the interactive prompts (if in camera mode):

• Webcam window will open
• Press SPACE to capture car images
• Press ENTER when finished capturing
• Press ESC to abort

Wait for processing.

Output locations:

• Final Article: Code/article.pdf
• Captured images: Code/webcam/webcam_images/
• Processed images: Code/webcam/processed_images/