Edge Developer Toolbox Developer Guide
Import a Custom AI Model
In the workbench tab, click the Import Custom Model tab.
Then choose your domain, and note that currently, the platform supports Computer Vision models, with plans to expand to Natural Language Processing models in the future.
Select the framework for your model, from the options TensorFlow, PyTorch, ONNX, and OpenVINO.
The steps you take after selecting the framework will vary based on the chosen framework; details such as model and file format will change according to your selections. If additional information is required, it will be requested at that step, ensuring that the tool adapts to your choices seamlessly.
For this example, we are importing our model YOLOv8 which was developed using the PyTorch* framework. After selecting Computer Vision as a domain and PyTorch as a framework, upload your AI model in the pt or pth file format to the designated directory. Provide a name for your model to easily identify it within the platform, aiding not only in distinguishing different models but also in keeping track of various preferences and versions throughout the optimization process. Click Submit to continue:
After a few moments, the upload is done, and you will see a “All files got uploaded to user directory” message. Click Next to continue: