Deploy an Arduino App Lab application to SOM-SMARC-Dragonwing-IQ8
Overview
This procedure moves an Arduino App Lab application from Arduino VENTUNO Q to SECO SOM-SMARC-Dragonwing-IQ8 running on the EVAL-KIT-SOM-SMARC with Carrier-SMARC-E71.
The example uses Detect Objects on Camera. The application is first checked on VENTUNO Q, exported as a ZIP package, transferred to the IQ8 target through Clea, then imported and started with Arduino App CLI.
The E71 evaluation kit is used as the hardware platform for the IQ8 SOM during bring-up and application validation.
What you need in advance
Prepare the following items:
- Arduino VENTUNO Q
- a USB camera supported by the App Lab example
- a development computer with Arduino App Lab
- SOM-SMARC-Dragonwing-IQ8
- SECO EVAL-KIT-SOM-SMARC with Carrier-SMARC-E71
- a Clea OS image for the target with Arduino App CLI available
- access to a Clea tenant
- the IQ8 evaluation device already registered in Clea Edgehog
- network connectivity for VENTUNO Q, the IQ8 evaluation kit, and the development computer
Before starting the application transfer, boot the IQ8 SOM on the E71 kit and confirm that the device is connected in Clea.
App Lab UI
The workflow below uses the
Inspirationssection available in the App Lab build used for this procedure. Recent builds can also expose the same application underExamples.
1. Run the reference application on VENTUNO Q
Open Arduino App Lab and connect to VENTUNO Q.
From the left sidebar, open:
Learn and Explore > Inspirations
Use the search field and enter:
object
Open Detect Objects on Camera.
The application uses a USB camera, the Video Object Detection Brick, and a WebUI component. Open the project README and check the camera requirements before starting the app.
Connect the USB camera to VENTUNO Q.
If App Lab asks for an execution mode, use Network Mode for this example.
Click Run in the upper-right corner.
Wait for the application to start. The web interface should open in the browser. Check that:
- the camera stream is visible;
- bounding boxes appear around detected objects;
- labels and confidence values are displayed;
- the web interface remains responsive during inference.
Keep this run as the reference for the IQ8 validation.
2. Export the application from App Lab
Keep Detect Objects on Camera open.
Click the arrow next to the application name in the upper-left area of App Lab.
Select:
Export App
The Export app dialog opens.
The dialog contains the option:
Include data associated with the App
Enable it only when the application data directory contains files that are required on the target.
Click:
Export .zip
Choose a local folder and save the exported application.
Keep the ZIP unchanged for the first deployment to the IQ8 target. This gives you the same application artifact that was already checked on VENTUNO Q.
3. Open the IQ8 evaluation device in Clea
Sign in to the Clea Portal.
From the left navigation, open:
Clea Edgehog > Fleet
Select the device associated with SOM-SMARC-Dragonwing-IQ8 on the E71 evaluation kit.
On the device page, check the connection state. The device should be shown as connected before starting the file transfer.
The device page exposes the runtime and management tabs used later in the procedure, including File Management and Applications.
4. Transfer the App Lab ZIP to the IQ8 target
On the IQ8 device page, open:
File Management
In File Management Mode, select:
Download to Device - Direct File
Set Destination to:
File System
Enter the destination path for the ZIP package on the target filesystem.
Example:
/home/arduino/Detect-Objects-on-Camera.zip
Choose the application ZIP exported from App Lab.
Keep Encoding set to None for a standard ZIP transfer.
Submit the file-management request.
At the bottom of the page, open Request History and follow the transfer until Clea reports a completed state.
The ZIP is now available on the IQ8 target at the path entered in the request.
5. Import the application with Arduino App CLI
Open a shell on the IQ8 target. You can use the remote terminal available from the Clea device page or another shell connection already configured for the board.
Run:
arduino-app-cli app import /home/arduino/Detect-Objects-on-Camera.zip
If you used a different destination path in File Management, use that path in the command.
List the applications available on the target:
arduino-app-cli app list
Locate Detect Objects on Camera and copy its App ID.
Start the application:
arduino-app-cli app start <APP_ID>
Replace <APP_ID> with the value returned by arduino-app-cli app list.
To inspect the application output from the terminal, run:
arduino-app-cli app logs <APP_ID>
Add --follow if supported by the App CLI version installed on the image and you want to keep the log stream open.
6. Check camera access on the E71 evaluation setup
Connect the camera selected for the IQ8 validation.
The physical connection can differ from the VENTUNO Q setup. Check the Linux device exposed to the application and the device permissions used by the runtime.
Start the app and compare the result with the VENTUNO Q reference.
Check:
- camera initialization;
- capture resolution;
- object classes returned by the model;
- confidence values;
- application UI;
- inference stability;
- application logs.
A different camera can produce different framing, color response, exposure, and timing. Record these differences separately from application-level behavior.
7. Restart the application
Stop the application:
arduino-app-cli app stop <APP_ID>
Start it again:
arduino-app-cli app start <APP_ID>
Check that the camera and inference pipeline return to the same working state.
This restart test catches missing runtime dependencies and device mappings that can remain hidden during the first launch.
Result
The same App Lab application has now been:
- executed on Arduino VENTUNO Q;
- exported as an App Lab ZIP;
- transferred through Clea;
- imported on SOM-SMARC-Dragonwing-IQ8;
- started through Arduino App CLI;
- checked on the E71 evaluation platform.
The E71 setup provides the reference hardware environment for the SOM. I/O mapping, production camera integration, power, thermal design, and carrier-board customization can be addressed after this application-level validation.
Compatibility
The Clea transfer and deployment workflow can also be used with Qualcomm-based products supported by Clea OS. Check the target BSP, camera path, device mappings, and accelerator support for the selected product.