Data Management
---------------
.. role:: guimenu

.. role:: guiaction

.. role:: guioption

.. contents::
    :local:
    :depth: 2
    :backlinks: none
    :class: toc

Data Management Sub-Tab
~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the :guimenu:`DATA MANAGEMENT` sub-tab, you have full control over your
mesh models, datasets, and trained AI models. This section allows you to
efficiently organize and manage your assets by providing options to list,
search, inspect, rename, and delete them as needed.


.. figure:: /_static/ai_hub/ai-hub-menu.png
    :align: center
    :class: padded-image

Mesh Models
++++++++++++++

All objects in the database are displayed in :guimenu:`Mesh Models`, along with their file name, size and creation date.
Using the actions on the right, you can either preview, delete or rename an object.
Click the :strong:`Play` button to preview the 3D object.

.. figure::  /_static/ai_hub/data_management/ai-hub-visualize-object.jpg
    :align: center
    :class: padded-image


Datasets
++++++++++++++

In the :guimenu:`Datasets` section, the datasets available can be managed. Click the :strong:`Play` button on the right to 
get detailed information about the dataset. E.g.,  object class names, dataset split sizes and a sample image from each data type are displayed.

.. figure::  /_static/ai_hub/data_management/ai-hub-manage-datasets.jpg
    :align: center
    :class: padded-image


AI Models
++++++++++++++

Manage pre-trained models and it's versions in the
:guimenu:`Trained Segmentation Models` and 
:guimenu:`Trained Pose Estimation Models` sections. Click on the `Extend` button on
the right to get detailed information on the model, e.g., datasets used to
train it, objects it can detect etc.

.. hint::
    A new version of an existing model becomes available when a new model with
    the similar name, dataset(s) and object(s) is trained.

.. figure::  /_static/ai_hub/data_management/ai-hub-manage-model.jpg
    :align: center
    :class: padded-image

Data Transfer Sub-Tab
~~~~~~~~~~~~~~~~~~~~~~~
The :guimenu:`DATA TRANSFER` sub-tab allows you to upload custom object meshes
and bin meshes to the database from an external drive. The transfer pipeline
would extract key attributes of objects within the uploaded mesh file(s),
process them, and store them in the database for various tasks across the
lifecycle of a robot application.

.. note::
    See :doc:`Tutorial: Upload A Mesh Model </tutorials/upload_mesh>`

Mesh model file requirements
++++++++++++++++++++++++++++++

.. list-table::
    :widths: 20 40 40
    :header-rows: 1

    * - Requirement
      - Description
      - Action to take if requirement is not met
    * - File name
      - Match the object name to the mesh file name or the zip file name if the mesh is zipped in a directory.
      - None
    * - Extension
      - The mesh file extension is either ``.obj`` or ``.ply``
      - Convert the file to ``.obj`` or ``.ply`` format using an open-source tool like Blender or MeshLab
    * - Format
      - The mesh file format is ASCII
      - Export the file as ASCII format using an open-source tool like Blender or MeshLab
    * - Scale
      - The mesh is scaled into millimeters
      - None
    * - Completeness
      - All faces are triangulated
      - Triangulate the mesh using an open-source tool like Blender or MeshLab
    * - Axes
      - The object is aligned w.r.t. principal axes, with Z-direction pointing upwards
      - None
    * - Texture
      - Texture is provided as ``.png`` file with the same file name as the mesh file
      - None
    * - Color information
      - Color information is stored in vertices not faces of the mesh
      - Transfer color information from face to vertex using an open-source tool like Blender or MeshLab
    * - Bins Shape
      - Rectangular-shaped
      - None
    * - Bins Orientation
      - Opening is pointing in +Z direction
      - None