GLOBALISE Loghi Handwritten Text Recognition Model - August 2023

This is a Loghi Handwritten Text Recognition (HTR) model that was created to transcribe text on the scans of the GLOBALISE VOC corpus. It was finetuned using Ground Truth that is also available in this Dataverse (GLOBALISE Ground Truth for Handwritten Text and Layout Recognition) and applied using the Loghi HTR tools to generate VOC transcriptions v2 - GLOBALISE. The model's hyperparameters are available in the file metadata.json.

Distributies

JSON
JSON
metadata.json
https://datasets.iisg.amsterdam/api/access/datafile/14478
Bestandsgrootte: 53.3 KB
application/octet-stream
saved_model.pb
https://datasets.iisg.amsterdam/api/access/datafile/14475
Bestandsgrootte: 15.4 MB
application/octet-stream
keras_metadata.pb
https://datasets.iisg.amsterdam/api/access/datafile/14481
Bestandsgrootte: 400.6 KB
application/octet-stream
variables.index
https://datasets.iisg.amsterdam/api/access/datafile/14479
Bestandsgrootte: 42.3 KB
application/octet-stream
variables.data-00000-of-00001
https://datasets.iisg.amsterdam/api/access/datafile/14477
Bestandsgrootte: 383.1 MB
text/plain
application/octet-stream

Registratie

Geregistreerd op
15 april 2024 om 12:39
Laatst gelezen
19 uur geleden
Kwaliteitsscore
100%