Deploying Neural Network to your application
Export the trained Neural Network
In order to use the "trained neural network" in your application, it has to be exported into a special file that contains its weights and structure information. You need to export the file in the format that matches your application.
For example, if you are deploying your training model in the LabVIEW application, then you need to export the file in LabVIEW supported format.
A supported application programming interface (API) will be provided to integrate a trained deep learning model file in your application. ANSCENTER provides deep learning APIs for LabVIEW, C++, and C# programming languages. Other programming languages will be supported in the next releases.
Figure 2.17: Deploying trained Neural Network into different environments
Once you've done your evaluation of your model, you can further test your model and visualize result before exporting the network file.
- Load your Test data-set by using the browse button.
- The data-set will be listed in the Test Files table.
- Manual click the file of interest to test, or use the navigation button to iterate through the loaded files.
- You may choose to show the image in its original size by clicking the Original Size tick-box.
- The predicted outcome for the selected test file will be shown on the right, with the raw calculated score for each class.
- Once you are satisfied with the quality of the model, choose the type of model to export. The exported model can be loaded using associated programming APIs, provided by the ANSCENTER.
The options are:
- Press EXPORT. A popup browser will allow you to name and save your exported model.
Notes: ANSCENTER is continuously working on supporting new export options. These options will be introduced in the next releases.