Model Resource

Definition

You can teach our system to recognize your custom concepts. This knowledge is stored in a Model and can be used in a Visual Mining Job for these modules: Face Recognition, Object & Scene Recognition and Landmark Recognition.

There are pre-trained Models ready to use and you can train your own custom model.

ENDPOINTS

GET /v1/models/

GET /v1/models/{MODEL_ID}/

PUT /v1/models/{MODEL_ID}/

DELETE /v1/models/{MODEL_ID}/

Pre-trained Models

Model	Visual Mining Module	Description
celebrities	Face Recognition	20.000 personalities, including the world's most famous people and a vast majority of German politicians and athletes
general-c	Object & Scene Recognition	Various objects and scenes, from general to more specific ones
zero-shot	Object & Scene Recognition	Recognize new unseen labels provided via dictionaries together with a set of preset labels
general-b	Landmark Recognition	Important sights, architectural structures and natural monuments across Europe + North America
europe	Landmark Recognition	Important sights, architectural structures and natural monuments only across Europe

Attributes

Name	Type	Description
id	string	Global identifier to access the actual resource
name	string	Name of the model
description	string	Description or notes of the model
versions	List of ModelVersion objects	A list of all versions (new version is added on training)
latest_version	ModelVersion object	The latest version as ModelVersion object
type	string	Type of the model (corresponding to the type of the dataset that was used during training)
is_public	boolean	Whether the model is pre-trained by DeepVA (public) or custom trained
time_created	string	Creation time of the dataset (ISO Time String)
time_updated	string	Modification time of the dataset (ISO Time String)
custom_fields	List of dict	Custom data can be bound to the model

JSON Example

The following JSON snippet is showing a custom trained Model object.

{
  "id": "90dc4c05-9989-4fa6-a1c7-702c7eea8bcb",
  "name": "DeepVA Team",
  "description": "Recognizing the DeepVA Team",
  "type": "face",
  "is_public": false,
  "time_created": "2020-02-27 17:46:44.006000",
  "time_updated": "2020-02-27 17:46:44.006000",
  "versions": [
    {
      "version_number": 1,
      "time_created": "2020-01-01 00:00:00.786000",
      "changelog": "Initial training",
      "source_dataset": {
        "id": "812ad5e6-5e65-4de4-a02a-0484072047b0",
        "name": "Team",
        "description": "This is the DeepVA team dataset",
        "type": "face",
        "time_created": "2020-02-20 17:54:06.854000",
        "time_updated": "2020-02-20 17:54:06.854000",
        "number_of_classes": 7,
        "number_of_active_classes": 7,
        "number_of_images": 4,
        "number_of_active_images": 4,
        "preview_images": [
          "storage://JL5Uh6yZFxdRO7DEdnCY"
        ],
        "evaluation_result": {}
      }
    }
  ],
  "latest_version": {
      "version_number": 1,
      "time_created": "2020-01-01 00:00:00.786000",
      "changelog": "Initial training",
      "source_dataset": {
        "id": "812ad5e6-5e65-4de4-a02a-0484072047b0",
        "name": "Team",
        "description": "This is the DeepVA team dataset",
        "type": "face",
        "time_created": "2020-02-20 17:54:06.854000",
        "time_updated": "2020-02-20 17:54:06.854000",
        "number_of_classes": 7,
        "number_of_active_classes": 7,
        "number_of_images": 4,
        "number_of_active_images": 4,
        "preview_images": [
          "storage://JL5Uh6yZFxdRO7DEdnCY"
        ],
        "evaluation_result": {}
      }
  },
  "custom_fields": []
}