Requirements
📦 Shared Requirements Architecture
All PyPottery tools are designed to work independently, yet they share a unified set of dependencies. This modular approach means you only need to install once and can use any combination of tools seamlessly.
PyPotteryInk
PyPotteryLens
PyPotteryLayout
Core Technologies
The PyPottery ecosystem is built on these foundational technologies:
PyTorch
v2.9.0+cu126The deep learning framework powering all AI capabilities. Includes CUDA 12.6 support for GPU acceleration.
- Neural network training & inference
- GPU-accelerated tensor operations
- TorchVision for image processing
Transformers
v4.57.1Hugging Face's library for state-of-the-art NLP and vision models. Powers OCR and text understanding.
- Pre-trained model access
- Text recognition (OCR)
- Few-shot learning capabilities
OpenCV
v4.12.0Computer vision library for image processing, manipulation, and analysis tasks.
- Image reading & writing
- Morphological operations
- Contour detection & analysis
Ultralytics YOLO
v8.3.223State-of-the-art object detection framework for identifying pottery drawings in documents.
- Real-time object detection
- Instance segmentation
- Custom model training
SAM2
v1.1.0Meta's Segment Anything Model for precise image segmentation and mask generation.
- Zero-shot segmentation
- Interactive mask editing
- High-quality boundaries
Diffusers
v0.35.2Hugging Face's diffusion models library for image enhancement and restoration.
- Image super-resolution
- Drawing enhancement
- Noise reduction
Data Processing
Pandas
v2.3.3Powerful data manipulation and analysis library for structured data handling.
- DataFrame operations
- Data cleaning & transformation
- Excel/CSV export
Polars
v1.35.1Lightning-fast DataFrame library for high-performance data processing.
- Parallel processing
- Memory efficient
- Lazy evaluation
PyMuPDF
v1.26.5High-performance PDF processing library for extracting images and text from documents.
- PDF page rendering
- Image extraction
- Text extraction
Visualization & Export
Matplotlib
v3.10.7Comprehensive plotting library for creating static, animated, and interactive visualizations.
- Publication-quality figures
- Image display & annotation
- Customizable layouts
ReportLab
v4.4.4PDF generation toolkit for creating professional publication layouts and plates.
- PDF document creation
- Vector graphics support
- Layout automation
SVGWrite
v1.4.3SVG generation library for creating scalable vector graphics from traced drawings.
- Vector output generation
- Path manipulation
- Scalable exports
System Requirements
- GPU: NVIDIA GPU with CUDA 12.6 support (8GB+ VRAM recommended)
- RAM: 16GB minimum, 32GB recommended for large batch processing
- Storage: SSD recommended for faster model loading
PyPottery requires Python 3.12. We recommend using a virtual environment to manage dependencies.