Test and Deploy
===============================

Deploying and Testing the Model in DaoAI World
-----------------------------------------------

    Once the model training is complete, you can test your model in DaoAI World.

Testing the Model on the Validation Set
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

    You can check the inform of inference time and GPU mode.

    .. image:: Images/inference_time_gpu.png
            :scale: 100%
            :align: center

|

    Select the model you want to test, then click the ``view test set``.

    .. image:: Images/test_set.png
        :scale: 70%
        :align: center

|

    Then you can see the images from the training set, validation set, and test set. In the "Discrepancy set," all images where the predictions differ from the annotations are also listed.

    .. image:: Images/test_set2.png
        :scale: 70%
        :align: center

|

    On the right side, you can toggle between the actual situation and the model predictions to compare the annotation details with the actual prediction results of the model.

    **Ground Truth**

    .. image:: Images/test_anno.png
        :scale: 60%
        :align: center

    **Model Prediction** (You can drag the confidence threshold slider on the left to filter out model predictions with low confidence levels.)

    .. image:: Images/test_label.png
        :scale: 60%
        :align: center



Upload an image to test the model
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

    On the project sidebar, select **Visualize** to enter the visualize page. At the top, choose the model you wish to deploy and test. Then, either click **Select File** or drag and drop the file into the upload image area to upload an image.

    .. image:: Images/deploy.png
        :width: 800
        :align: center

    After uploading, the model will perform inference on the uploaded image. Once completed, you can view the model's recognition results in the window. The top right corner will display information such as the number of objects and keypoints detected in the current image.

    Additionally, you can adjust the **Confidence Threshold** on the right to filter out recognition results with low confidence levels.


View annotations and detection results at the same time
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Click the ``Display both Ground Truth and Model Predictions`` button on top-left corner, user can switch to different display: Ground Truth and Model Predictions.

    .. image:: Images/compare_annotate.png
        :width: 800
        :align: center

Click the ``Change Label Style`` button, after changing color user can simultaneously compare the annotation results with the model's predictions.

    .. image:: Images/change_color.png
        :width: 800
        :align: center


Deploying Models in Other Applications
---------------------------------------------

    Once model training is complete, you can export the trained model to your local device and use it in other applications.

    .. image:: Images/export_model.png
        :width: 800
        :align: center

    The exported model files support both ``.dwm`` and ``.zip`` formats.

    Models exported from DaoAI World can also be used in other DaoAI platform software, such as **DaoAI InspecTRA** and **DaoAI VisionPilot**.

    - ``.dwm`` format is supported in DaoAI World version **2.24.6.0** and later. The corresponding versions for DaoAI InspecTRA and DaoAI VisionPilot are **2.24.6.0** and **2.24.5.0**, respectively.
    - ``.zip`` format is available in DaoAI World version **2.24.5.0** and earlier. The corresponding versions for DaoAI InspecTRA and DaoAI VisionPilot are **2.24.5.0** and **2.24.4.3**, respectively.


Model Deploy
------------------

Click on ``Deploy`` button to the model deploy page. You can easily deploy your model to cloud, locally, or on an edge device.

.. image:: Images/deploy_sdk.png
    :scale: 80%
    :align: center

Click on ``View Code`` button on Hosted Image Inference, you can see your trained models in differrent versions. You can use the example code to make inference.

.. image:: Images/click_check_code.png
    :scale: 80%
    :align: center

Click on ``Switch Model`` to switch from different versions, different version has different **trained_model_uid**.

.. image:: Images/check_code.png
    :scale: 80%
    :align: center

Inference result return as format of **dictionary**, you can read the detailed document from :ref:`Hosted Image Inference`.

Click on ``View Documentation`` will redirect to DWSDK chapter of this document.

.. image:: Images/open_sdk_doc.png
    :scale: 80%
    :align: center