Integrate a custom AI model in an application - Qualcomm Dragonwing Documentation

This section describes the available paths for deploying a custom model using Qualcomm SDKs. To use a custom model in an AI application, you can use the Qualcomm Intelligent Multimedia SDK or Qualcomm AI Runtime SDK.

**Bring your own model: Time taken to deploy a model** — Bring your own model: Time taken to deploy a model

Using Qualcomm IM SDK

SDK supported use case?	SDK supported model?	Number of AI Hub models available	Expected time to integrate the model in an application
Yes	Yes	84	15 minutes This is an advanced use case.

Using Qualcomm AI Runtime SDK

SDK supported use case?	SDK supported model?	Number of AI Hub models available	Expected time to integrate the model in an application
No constraints	No constraints	NA	~1 week This is an advanced use case.

The moderate and advanced paths assume understanding of GStreamer fundamentals, understanding of GStreamer pipelines, expertise in C/C++, and a Yocto build environment.

⌘I