Get started with our pre-trained models

If you don’t have the resources to train your own model, these pre-trained will provide a good sandbox enviornment for testing. Download a model load it to the program, and you can use that to generate metadata for your own images.

ptm-im2txt-incv3-mlib-cleaned-1m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is Inception v3
  • Trained on historical photographs from Marriott Library’s digital collection
  • Captions used in the training went through NLP cleaning and proper-nouns removed
  • Trained with 1 million steps

MD5 Hash: 59f4246b9e12ecbed1480ca6925febcf

ptm-im2txt-incv3-mlib-cleaned-3m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is Inception v3
  • Trained on historical photographs from Marriott Library’s digital collection
  • Captions used in the training went through NLP cleaning and proper-nouns removed
  • Trained with 3 million steps

MD5 Hash: b7725432034f005237e75a9b6b54f205

ptm-im2txt-incv3-mlib-uncleaned-1m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is Inception v3
  • Trained on historical photographs from Marriott Library’s digital collection
  • Captions used in the training were original captions with proper-nouns
  • Trained with 1 million steps

MD5 Hash: ead9c5c21a8767f50ab5d865671949d7

ptm-im2txt-incv3-mscoco-1m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is Inception v3
  • Trained on MS COCO data
  • Trained with 1 million steps

MD5 Hash: 496172c51d1f4d1b35a937e00355a60f

ptm-im2txt-incv3-mscoco-3m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is Inception v3
  • Trained on MS COCO data
  • Trained with 3 million steps

MD5 Hash: 99c21baa98bf683580eef3d00213378b

ptm-im2txt-mscoco1m-mlib-cleaned-2m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is the model trained with MS COCO data for 1 million steps
  • Trained on historical photographs from Marriott Library’s digital collection
  • Captions used in the training went through NLP cleaning and proper-nouns removed
  • Trained with 2 million steps

MD5 Hash: 3c28e62806b66bd4fd062d85ab238e3d

ptm-im2txt-mscoco3m-mlib-cleaned-1m

Uses Show and Tell model that is a deep neural network that learns how to describe the content of images.

  • starting point is the model trained with MS COCO data for 3 million steps
  • Trained on historical photographs from Marriott Library’s digital collection
  • Captions used in the training went through NLP cleaning and proper-nouns removed
  • Trained with 1 million steps

MD5 Hash: ff4ec4f78b97f24754e31cb808209e36

N/A

N/A

MD5 Hash: N/A