Skip to content

architecture.md

Latest commit

 

History

History
56 lines (42 loc) · 1.86 KB

architecture.md

File metadata and controls

56 lines (42 loc) · 1.86 KB

Architecture

Overview

python-cvpy provides higher-level Python interfaces for SAS Viya image analytics workflows. It sits between Python client code and SAS CAS-backed image processing capabilities.

At a high level, the architecture is:

Python application code
    │
    ├── cvpy
    │     ├── table abstractions
    │     ├── image conversion utilities
    │     ├── visualization helpers
    │     ├── annotation integration
    │     └── CAS tuning helpers
    │
    ├── SWAT client
    │
    └── SAS Viya / CAS
          ├── image action set
          └── biomedimage action set

Design goals

The package is designed to:

  • reduce low-level handling of SAS-hosted image data in Python
  • expose more convenient abstractions for image tables
  • support conversion into numpy-friendly forms
  • integrate visualization tools commonly used in Python workflows
  • support annotation-related workflows where external tooling is required

Major package areas

  • cvpy.annotation — annotation models and CVAT-oriented integration
  • cvpy.base — shared base types and cross-cutting abstractions
  • cvpy.image — natural image table support
  • cvpy.biomedimage — biomedical image table support
  • cvpy.utils — conversion and helper utilities
  • cvpy.visualization — display and rendering functions

Operational model

In most workflows:

  1. a user authenticates to SAS Viya through SWAT
  2. CAS tables holding image data are loaded or referenced
  3. cvpy APIs are used to fetch, transform, or visualize data
  4. optional downstream annotation or optimization utilities are applied

Scope

python-cvpy is not a replacement for SWAT or CAS action sets. Instead, it provides a Python-friendly layer over common image analytics workflows that would otherwise require more manual manipulation.