# Run sample applications using a host computer
You can run the sample application on the evaluation kit using a host computer with Windows^®^, Ubuntu, or mac^®^ operating system.
## Prerequisites
Before you begin, [set up the device.](https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-ug-set-up-the-device.html#iq9-ug-set-up-the-device)
## Run multimedia sample applications
The multimedia sample applications show use cases for camera, display, and video streams on the Dragonwing IQ-9075 device.
### Multicamera streaming or encoding (Dash cam)
The **gst-multi-camera-example** command-line application shows simultaneous streaming from two camera sensors on the Dragonwing IQ-9075 device. The application composes the camera feeds side-by-side and displays them on a screen, or encodes and stores the video streams to files.
The following figure shows the dash cam application workflow for composing and encoding use cases.
**Figure: Dash cam application workflow**
The following video provides an overview of Multicamera application.
**Video: Multicamera application**
**Example**
To start the application and view, store, and verify the encoder output, do the following in a SSH terminal:
1. View the sample application on the screen by running the following export command:
export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_DISPLAY=wayland-1
Copy to clipboard
2. View the `waylandsink` output by running the following command:
gst-multi-camera-example -o 0
Copy to clipboard
3. Store and verify the encoder output as follows:
1. Store the encoder output by running the following command:
gst-multi-camera-example -o 1
Copy to clipboard
The device stores the encoded files from camera 1 and camera 2 at `/opt/cam1_vid.mp4` and `/opt/cam2_vid.mp4` directories, respectively.
2. Pull the encoded files from the host computer by running the following command:
scp root@:/opt/cam1_vid.mp4
Copy to clipboard
**Parameters**
>
>
> - `` is the IP adddress of the target device.
> - `destination_directory` is the directory to which the encoded file must be pulled to.
> - `/opt/cam1_vid.mp4` is the sample encoded file you want to pull.
Note
When prompted for a password, enter `oelinux123` to authenticate the file transfer.
3. Play the encoder output using any media player that supports MP4 files.
**Other functions or commands**
- To stop the use case, select CTRL + C.
- To display the available help options, run the following command:
>
>
> gst-multi-camera-example --help
> Copy to clipboard
- To debug GStreamer, set the `GST_DEBUG` environment variable to the required level. This variable controls the GStreamer debug output.
>
>
> For example, to log all warnings, run the following command:
>
>
> export GST_DEBUG=2
> Copy to clipboard
### Multichannel video decode and compose (Video wall)
The **gst-concurrent-videoplay-composition** command-line application enables concurrent video decoding and playback of AVC-coded videos, composing them on a display for video wall applications. The application requires at least one input video file in MP4 format with the AVC codec. The application supports up to 32 streams.
The following figure shows the video wall application workflow for decoding and display use cases.
**Figure: Video wall application workflow**
The following video provides an overview of Multichannel decode and display application.
**Video: Multichannel decode and display application**
**Example**
To start the application and concurrent playback for four channels, do the following in an SSH terminal:
1. Transfer prerecorded or test videos that are in the AVC-encoded MP4 (H.264) format to your device by running the following command on the host computer:
scp root@:/opt/
Copy to clipboard
**Parameters**
>
>
> - `` is the filename of the videos. Ensure that you use this filename syntax for all video files.
> - `` is the IP adddress of the device.
> - `/opt/` is the destination directory.
Note
When prompted for a password, enter `oelinux123` to authenticate the file transfer.
2. View the sample application on the screen by running the following export command in SSH:
export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_DISPLAY=wayland-1
Copy to clipboard
3. Start concurrent playback for four channels by running the following command:
gst-concurrent-videoplay-composition -c 4 -i /opt/.mp4 -i /opt/.mp4 -i /opt/.mp4 -i /opt/.mp4
Copy to clipboard
**Parameters**
>
>
> - `-c`: specifies the number of streams to be decoded for composition. The value can be 2, 4, 8, 16, or 32. In the command, `-c` is `4`.
> - `-i`: specifies the absolute path to the input video file. In the command, the absolute paths to the four files are `/opt/.mp4`, `/opt/.mp4`, `/opt/.mp4`, and `/opt/.mp4`.
**Other functions or commands**
- To stop the use case, select CTRL + C.
- To display the available help options, run the following command:
>
>
> gst-concurrent-videoplay-composition --help
> Copy to clipboard
- To debug GStreamer, set the `GST_DEBUG` environment variable to the required level. This variable controls the GStreamer debug output.
>
>
> For example, to log all warnings, run the following command:
>
>
> export GST_DEBUG=2
> Copy to clipboard
## Run AI sample applications
AI sample applications show use cases for object detection and parallel inferencing on input streams from a camera, video file, or Real-Time Streaming Protocol (RTSP) stream on the Dragonwing IQ-9075 device.
The following workflow shows the high-level steps to run AI sample applications.
**Figure: Workflow for running AI sample applications**
**Download and transfer AI models and labels**
To run AI sample applications, download the required models and labels, and transfer them to the Dragonwing IQ-9075 device. You can download the AI models from [Qualcomm® AI Hub](https://aihub.qualcomm.com/iot/models) and labels from GitHub.
To run object detection, parallel inferencing, and multistream inferencing AI use cases, you need the following AI models and labels:
| Sample application | Models required |
| --- | --- |
| AI object detection | [yolox_quantized.tflite](https://aihub.qualcomm.com/iot/models/yolox?searchTerm=yolo*) |
| Parallel AI inference | [yolox_quantized.tflite](https://aihub.qualcomm.com/iot/models/yolox?searchTerm=yolo*) |
| Parallel AI inference | [Inception-v3](https://aihub.qualcomm.com/iot/models/inception_v3) |
| Parallel AI inference | [HRNetPose](https://aihub.qualcomm.com/iot/models/hrnet_pose) |
| Parallel AI inference | [DeepLabV3-Plus-MobileNet](https://aihub.qualcomm.com/iot/models/deeplabv3_plus_mobilenet) |
| Multistream inference | [yolox_quantized.tflite](https://aihub.qualcomm.com/iot/models/yolox?searchTerm=yolo*) |
| Multistream inference | [Inception-v3](https://aihub.qualcomm.com/iot/models/inception_v3) |
| | |
| | |
| | |
To download and transfer the AI models and labels for object detection, parallel inferencing, and multistream inferencing, do the following:
1. Download and run the automated script to get the required model and label files on the device by running the following commands:
cd /tmp/
Copy to clipboard
curl -L -O https://raw.githubusercontent.com/quic/sample-apps-for-qualcomm-linux/refs/heads/main/qualcomm-linux/scripts/download_artifacts.sh
Copy to clipboard
chmod +x download_artifacts.sh
Copy to clipboard
./download_artifacts.sh
Copy to clipboard
2. Update the `q_offset` and `q_scale` constants of the quantized LiteRT model in the JSON file. For instructions, see [Get the model constants](https://docs.qualcomm.com/bundle/publicresource/topics/80-70020-15B/integrate-ai-hub-models.html#get-the-model-constants).
3. Push the updated quantized LiteRT model files to the `/etc/models/` directory on your device by running the following command:
scp inception_v3_quantized.tflite root@:/etc/models/
Copy to clipboard
**Parameters**
>
>
> - `inception_v3_quantized.tflite`: The JSON file of quantized LiteRT model.
> - ``: The IP address of the target device.
> - `/etc/models/`: The destination directory on your device.
4. Ensure that you have test videos on your device.
The `download_artifacts.sh` script downloads some test videos to the `/etc/media/` directory.
To push your own test videos to the device, run the following command from your host computer:
scp .mp4 root@:/etc/media/
Copy to clipboard
**Parameters**
>
>
> - `.mp4`: The filename of your MP4 test video.
> - ``: The IP address of the target device.
> - `/etc/media/`: The destination directory on your device.
You can now run AI sample applications.
### AI object detection
The **gst-ai-object-detection** sample application shows the hardware capability to detect objects on input streams from a camera, video file, or RTSP stream. The pipeline receives the input stream, preprocesses it, runs inferences on AI hardware, and displays the results on the screen.
The following figure shows the workflow for AI object detection application.
**Figure: Object detection application workflow**
The following video provides an overview of AI object detection application.
**Video: Object detection application**
**Example**
Before you begin running the sample application and its use cases, do the following:
>
>
> - Push the model and label files to the device as described in [Download and transfer AI models and labels](https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-run-apps-using-host.html#procedure-ai).
> - Ensure that the display monitor is on.
To start the AI object detection sample application and run its use cases, do the following:
1. Start a new SSH session by running the following command:
ssh root@
Copy to clipboard
**Parameters**
>
>
> `` is the IP address of the target device.
2. View the sample application on the screen by running the following export command in SSH:
export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_ DISPLAY=wayland-1
Copy to clipboard
3. Edit the `/etc/configs/config_detection.json` file on your device as follows:
{
"file-path": "/etc/media/video.mp4",
"ml-framework": "tflite",
"yolo-model-type": "yolox",
"model": "/etc/models/yolox_quantized.tflite",
"labels": "/etc/labels/yolox.labels",
"constants": "YOLOx,q-offsets=<38.0, 0.0, 0.0>,q-scales=<3.6124823093414307, 0.003626860911026597, 1.0>;",
"threshold": 40,
"runtime": "dsp"
}
Copy to clipboard
**Parameters**
The following table lists and describes the fields in the configuration file.
| Field | Values/description |
| --- | --- |
| **ml-framework** | **ml-framework** |
| `snpe` | Uses the Qualcomm^®^ Neural Processing SDK models. |
| `tflite` | Uses the LiteRT models. |
| `qnn` | Uses the Qualcomm^®^ AI Engine direct models. |
| **yolo-model-type** | **yolo-model-type** |
| `yolov5` `yolov8` `yolonas` `yolox` | Runs the YOLOv5, YOLOv8, YOLO-NAS, and YOLOx models, respectively. See [Sample model and label files](https://docs.qualcomm.com/bundle/publicresource/topics/80-70020-50/gst-ai-object-detection.html#gst-ai-object-detection__section_hds_vxp_mdc). |
| **runtime** | **runtime** |
| `cpu` | Runs on the CPU |
| `gpu` | Runs on the GPU |
| `dsp` | Runs on the digital signal processor (DSP) |
| **Input source** | **Input source** |
| `camera` |
0 – Primary camera
1 – Secondary camera
|
| `file-path` | Directory path to the video file |
| `rtsp-ip-port` | Address of the RTSP stream in `rtsp://:/` format |
4. Start the `gst-ai-object-detection` application by running the following command:
gst-ai-object-detection
Copy to clipboard
**Other functions or commands**
- To stop the use case, select CTRL + C.
- To display the available help options, run the following command:
>
>
> gst-ai-object-detection -h
> Copy to clipboard
- To debug GStreamer, set the `GST_DEBUG` environment variable to the required level. This variable controls the GStreamer debug output.
>
>
> For example, to log all warnings, run the following command:
>
>
> export GST_DEBUG=2
> Copy to clipboard
### Parallel AI inference
The **gst-ai-parallel-inference** command-line application shows the hardware capability to perform four parallel AI inferences on input streams from a camera, video file, or RTSP stream. The pipeline detects objects, classifies objects, detects poses, and segments images on the input stream. The screen displays the results side-by-side.
The following figure shows the workflow for parallel AI inference application.
**Figure: Parallel inference application workflow**
The following video provides an overview of parallel AI inference application.
**Video: Parallel inferencing application**
**Example**
Before you begin running the sample application and its use cases, do the following:
>
>
> - Push the model and label files to the device as described in [Download and transfer AI models and labels](https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-run-apps-using-host.html#procedure-ai).
> - Ensure that the display monitor is on.
To start the parallel AI inference sample application and run its use cases, do the following:
1. Start a new SSH session by running the following command:
ssh root@
Copy to clipboard
**Parameters**
>
>
> `` is the IP address of the target device.
2. View the sample application on the screen by running the following export command in SSH:
export XDG_RUNTIME_DIR=/dev/socket/weston && export WAYLAND_ DISPLAY=wayland-1
Copy to clipboard
3. Edit the `/etc/configs/config-parallel-inference.json` file on your device as follows:
{
"file-path": "/etc/media/video.mp4",
"detection-model": "/etc/models/yolox_quantized.tflite",
"detection-labels": "/etc/labels/yolox.labels",
"detection-constants": "Yolox,q-offsets=<38.0, 0.0, 0.0>,q-scales=<3.6124823093414307, 0.003626860911026597, 1.0>;",
"pose-model": "/etc/models/hrnet_pose_quantized.tflite",
"pose-labels": "/etc/labels/hrnet_pose.labels",
"pose-constants" : "Posenet,q-offsets=<8.0>,q-scales=<0.0040499246679246426>;",
"segmentation-model": "/etc/models/deeplabv3_plus_mobilenet_quantized.tflite",
"segmentation-labels": "/etc/labels/deeplabv3_resnet50.labels",
"segmentation-constants": "deeplab,q-offsets=<0.0>,q-scales=<1.0>;",
"classification-model": "/etc/models/inception_v3_quantized.tflite",
"classification-labels": "/etc/labels/classification.labels",
"classification-constants": "Inceptionv3,q-offsets=<38.0>,q-scales=<0.17039915919303894>;"
}
Copy to clipboard
**Parameters**
The following table lists and describes the fields in the file to configure the input source for the sample application.
| **Field** | **Values/description** |
| --- | --- |
| **Input source** | **Input source** |
| `camera` |
0 – Primary camera
1 – Secondary camera
|
| `file-path` | Directory path to the video file |
| `rtsp-ip-port` | Address of the RTSP stream in `rtsp://:/` format. |
4. Start the `gst-ai-parallel-inference` application by running the following command:
gst-ai-parallel-inference
Copy to clipboard
**Other functions or commands**
- To stop the use case, select CTRL + C.
- To display the available help options, run the following command:
>
>
> gst-ai-parallel-inference -h
> Copy to clipboard
- To debug GStreamer, set the `GST_DEBUG` environment variable to the required level. This variable controls the GStreamer debug output.
>
>
> For example, to log all warnings, run the following command:
>
>
> export GST_DEBUG=2
> Copy to clipboard
**Known issue**
In pose detection, the model detects only one person, even if many people are present in the frame.
Note
Image classification using the Inception v3 model trains on the ImageNet data set. As a result, the model can’t detect a person because this class isn’t included in the data set.
Note
Qualcomm AI Hub often updates models with the latest SDK versions. Using wrong model constants may lead to inaccurate results. If you face such issues, update the model constants in the configuration file.
### Multistream inference
The **gst-ai-multistream-inference** application shows AI inference (object detection and classification) on up to 32 input streams coming from a camera, file, or RTSP stream. The pipeline processes multiple input streams, performs preprocessing and AI inferencing, combines the streams, and merges them into a single video output. The output is displayed on the screen, saved as an H.264 encoded MP4 file, or converted into an RTSP stream.
The following figure shows the workflow for multistream inference application.
**Figure: Multistream inference application workflow**
**Example**
Before you begin running the sample application and its use cases, do the following:
>
>
> - Push the model and label files to the device as described in [Download and transfer AI models and labels](https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-run-apps-using-host.html#procedure-ai).
> - Ensure that the display monitor is on.
To start the multistream inference sample application and run its use cases, do the following:
1. Start a new SSH session by running the following command:
ssh root@
Copy to clipboard
**Parameters**
>
>
> `` is the IP address of the target device.
2. Edit the `/etc/configs/config-multistream-inference.json` file as follows:
{
"input-file-path":
[
"/etc/media/video1.mp4",
"/etc/media/video2.mp4",
"/etc/media/video3.mp4",
"/etc/media/video4.mp4",
"/etc/media/video5.mp4",
"/etc/media/video6.mp4",
"/etc/media/video7.mp4",
"/etc/media/video8.mp4",
"/etc/media/video9.mp4",
"/etc/media/video10.mp4",
"/etc/media/video11.mp4",
"/etc/media/video12.mp4",
"/etc/media/video13.mp4",
"/etc/media/video14.mp4",
"/etc/media/video15.mp4",
"/etc/media/video16.mp4"
],
"model": "/etc/models/yolox_quantized.tflite",
"labels": "/etc/labels/yolox.labels",
"input-type": "h264",
"output-display": 1,
"use-case": 0
}
Copy to clipboard
**Parameters**
The following table lists and describes the fields in the configuration file.
| **Field** | **Values/description** |
| --- | --- |
| **Input sources** | **Input sources** |
| `num-camera` | Number of camera inputs. Maximum value: 2 |
| `camera-id` | Valid only if `num-camera=1`
0: Primary camera
1: Secondary camera
|
| `input-file-path` | JSON array of input files. For example, `["/etc/media/video1.mp4", "/etc/media/video2.mp4", ....]` |
| `input-rtsp-path` | JSON array of rtsp video streams For example, `["`rtsp://192.168.1.4:8554/live1.mkv `_", "`rtsp://192.168.1.4:8554/live2.mkv `_", ....]` |
| **Input type** | **Input type** |
| `h264` | Input is H264 encoded |
| `h265` | Input is H265 encoded |
| **Use case** | **Use case** |
| `0` | Object detection |
| `1` | Classification |
| **Output options** | **Output options** |
| `output-file-path` | Path to output file |
| `output-ip-address` | IP address for output RTSP stream |
| `output-port-number` | Port number of output RTSP stream |
| `output-display` |
1: Enable display output
2: Disable display output
|
3. Start the `gst-ai-multistream-inference` application by running the following command:
gst-ai-multistream-inference
Copy to clipboard
Last Published: May 21, 2026
[Previous Topic
Use the Qualcomm Visual Studio Code Extension](https://docs.qualcomm.com/bundle/publicresource/80-70020-261/topics/iq9-run-apps-using-ide.md) [Next Topic
Develop an application](https://docs.qualcomm.com/bundle/publicresource/80-70020-261/topics/iq9-ug-develop-an-app.md)
Source: [https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-run-apps-using-host.html](https://docs.qualcomm.com/doc/80-70020-261/topic/iq9-run-apps-using-host.html)