It helps users quickly understand, explore, and prototype ideas by showcasing the platform’s performance and capabilities—inspiring innovation through hands-on experience.
Note
iQ-Studio is suitable for a range of products developed based on the Innodisk Qualcomm Dragonwing SoC.
iQ-Studio is built upon a robust edge AI software stack, bridging the gap between hardware and high-level applications:
- Hardware & Firmware: Qualcomm Dragonwing QCS9075 SoC and low-level firmware.
- Kernel Space: Powered by Qualcomm Linux, integrated with our custom Inno DTB/drivers and Yocto environments.
- User Space: Seamlessly supports 3rd-party LLM SDKs, device management (iCAP), and inno AVL. At the very top sits the iQS-App layer (VLM, Streampipe, YOLO, OGenie).
Our architecture evolves alongside the Qualcomm Linux Roadmap, ensuring alignment with the latest kernel and Yocto Project releases:
| Linux Kernel | Yocto Project | Qualcomm Linux (QLI) Release |
|---|---|---|
| 6.6 LTS | 4.0 Kirkstone | QLI 1.x |
| 6.6 LTS | 5.0 Scarthgap | QLI 1.x |
| 6.18 LTS | Wrynose (Master) | QLI 2.x |
We ensure a continuous and stable pipeline—from upstream Linux/Yocto projects down to the optimized downstream drivers—unlocking peak performance for edge AI workloads.
We also provide integrated and supplied Ubuntu images for development.
Before getting started, please refer to the Starting Guides to boot up your platform. As with the Q911 series, please refer to the EXMP-Q911 Starting Guide.
iQ Studio enables users to run applications quickly. It supports both online and offline modes, ensuring that applications can run even without internet access. Currently, we provide two types of application packages—docker images and IPK packages—that can be executed with IQ Studio.
If you are using online mode (with internet access), you only need to install the iQ studio github repository on the platform and run our applications by following the tutorial commands. For usage instructions, please refer to the Quick Start guide.
If you must use offline mode (without internet access), you need to first transfer the required packages and the iQ studio github repository to the platform before you can run the applications in an offline environment. For usage instructions, please refer to the how to install offline package.
We verify the BSP version on the platform to ensure that applications run correctly. This check is automatically handled by iqs-launcher. However, it is important to confirm that the BSP version on your system matches the one specified in the tutorial before running any application.
git clone https://github.com/InnoIPA/iQ-Studio.git
cd iQ-Studio
./install.shNote: If you are using Ubuntu, please log in again after installation.
For example, If you want to run the iQ-VLM. You only need two command run the interative real-time demo.
Launch the OGenie API server:
iqs-launcher --autotag iqs-ogenieReal-Time Display of VLM Predictions on the Monitor.
iqs-launcher --autotag iqs-vlm-demoThis provides a real-time display of VLM predictions, allowing you to quickly verify the inference results.
For other applications, please refer to the documentation section below.
iQ Studio resources are grouped into categories based on functionality:
| Categories | Description | Topic |
|---|---|---|
| Starting Guides | Quick-start guides for everything. | |
| AVL(Approved Vendor List) | Provides guidance on verifying that the driver starts correctly on the system and quickly demonstrating the validated results. | |
| Applications | Application-level examples focused on specific use cases and vertical scenarios. | |
| Model Deploy | End-to-end guides for turning trained AI models into target-ready artifacts, covering quantization, conversion, quality validation, and on-device inference. | |
| SDKs | Documentation and examples on how to use the SDKs effectively. | |
| Benchmarks | Performance tests and comparisons across platforms. |
Please refer to the Changelog for all updates.
This project is licensed under the MIT License. See LICENSE for details.