Testing the Transkribus API

The aim of this repo is to keep track of my attempts to learn how to interface with the API for Transkribus using primarily Python.

This attempt is first to learn how to interact with API, then to interact with Transkribus

Transkribus is an AI platform which has a focus on OCR tasks.

It describes itself in the followig manner:

Transkribus is an AI-powered platform for text recognition, transcription and searching of historical documents – from any place, any time, and in any language.

Transkribus | AI powered Handwritten Text Recognition

Their API documentation is a set of Restful methods to facilitate communication between the Transkribus application and the server. It can be summarised in the following manner:

• Login and Authentication: The user has to POST their credentials to the login method and include the Session-ID in subsequent requests.

• Collections and Documents: The user can access, list, and retrieve the collections and documents they have rights to via GET requests.

• Jobs and Processing Tasks: The user can start, monitor, cancel, and query the details and errors of various processing tasks, such as document creation, layout analysis, HTR, OCR, etc., via POST and GET requests.

• Search and Recognition: The user can perform full-text search and OCR recognition on documents and pages via GET and POST requests.

Our intitial goal will be to do the following:

• Authenticate
• Upload an image
• Kick off a transcription job (you can do that through the web interface)
• Download the completed text transcription of the image