LayoutParser PDF Layout Processor

This project provides a script to sample random pages from PDF files, detect layout zones using LayoutParser (with an optional Detectron2 backend on CPU via WSL), visualize the results, and save annotated images.

---

Prerequisites

• Windows 10/11 with WSL 2 (Ubuntu 20.04+)
• Python 3 (≥ 3.8)
• pip, virtualenv, and system packages (poppler-utils, build-essential, cmake, etc.)

---

Setup

1. Clone this repository

cd /mnt/c/Users/nikdu/my_python_projects
# Replace <repo-url> with your repo or local path
git clone <repo-url> layout_parser_linux
cd layout_parser_linux

2. Create & activate a virtual environment

# In your Ubuntu shell (WSL)
python3 -m venv .venv
source .venv/bin/activate

3. Install system dependencies

sudo apt update
sudo apt install -y poppler-utils build-essential cmake pkg-config libjpeg-dev libpng-dev

4. Install Python dependencies

pip install --upgrade pip setuptools wheel cython
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
pip install 'git+https://github.com/facebookresearch/detectron2.git@v0.6'
pip install layoutparser pdf2image PyPDF2 matplotlib

5. Verify installation (optional)

python - <<'EOF'
import torch, detectron2, layoutparser as lp
print("torch:", torch.__version__)
print("detectron2:", detectron2.__version__)
print("layoutparser:", lp.__version__)
EOF

---

Usage

1. Prepare your PDFs: place PDF files in a folder, e.g.: /mnt/c/Users/nikdu/my_python_projects/layout_parser_linux/pdfs

2. Configure paths in the script or notebook cells:

   • folder_path: input PDF directory
   • output_folder: directory to save annotated images

3. Run the processor:

python layout_test.py

By default, the script samples up to 10 random pages per PDF, detects layout zones, displays them inline (if enabled), and saves PNGs to your output folder.

---

Project Structure

layout_parser_linux/
├── .venv/                 # Python virtual environment
├── layout_test.py         # Main processing script
├── layoutprocessor.ipynb  # (Optional) Jupyter notebook version
├── pdfs/                  # Place your PDF files here
├── output/                # Annotated images will be saved here
└── README.md              # This file

---

Notes

• CPU-only: This setup uses a CPU‑only Detectron2 build. For GPU support, use a CUDA‑enabled environment or Docker as previously described.

• Parameters: Adjust sample_size, dpi, and display_inline in the script to suit your needs.

• Reproducibility: To share your exact environment, run:

  pip freeze > requirements.txt

Happy layout extracting 🚀