CarpeDM: Sieze the Data Manager

Description

CarpeDM is a general library for downloading, viewing, and manipulating image data. Originally developed as a ChARacter shaPE Data Manager, CarpeDM aims to make Japanese character shape (字形) data and other image datasets more accessible to machine learning researchers.

Datasets Currently Available for Download

ID	Dataset
pmjtc	Pre-Modern Japanese Text Character Shapes Dataset (日本古典籍字形データセット), provided by the Center for Open Data in the Humanities (CODH).

Though still in the early stages of development, a high-level interface is also provided for

• Automatic model-ready data generation.
• Flexible training of models with a variety of deep learning frameworks.

Currently supported deep learning frameworks:

• TensorFlow

Documentation

• Installation
  • Recommended Environments
  • Dependencies
  • Install CarpeDM
  • Uninstall CarpeDM
  • Upgrade CarpeDM
  • FAQ
• Guides
  • Basic Usage
• Examples
  • Single Character Task
  • Baseline Model
  • Using Tasks and Models
• API
  • Data
  • Neural Networks
  • Models
  • Tasks
  • Utilities

Other

• Benchmarks
• Contributing
• Code of Conduct
• License

Indices and tables

• Index
• Module Index
• Search Page