> ## Documentation Index
> Fetch the complete documentation index at: https://dragonwingdocs.qualcomm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# End-to-end samples

> Run AI samples for image classification and hand detection with published input images. No camera is required for these samples.

Run the QIR SDK AI vision samples end-to-end without any camera. Both samples use `image_publisher` to publish a bundled image at a fixed rate, then push the frame through the same QNN inference pipeline as the camera launch variant. The samples run on the IQ-9075 EVK under either Qualcomm Linux or Ubuntu, and on the Qualcomm Dragonwing™ RB3 Gen 2 Vision Kit under Ubuntu.

## Prerequisites

Confirm the following before you start either sample.

* For more information, see the Evaluate tab on the [Scenarios](./scenarios#evaluate) page.
* The host and device can reach each other over the network using the `ping` command.
* ROS 2 Jazzy is available on both the host and the device with a matching `ROS_DOMAIN_ID`.
* The device runs `qcom-robotics-proprietary-image` (Qualcomm Linux) or `qirp-sdk` or `qirp-sdk-desktop` (Ubuntu).

## Run AI samples

Follow the tab for the sample you want to run. Both samples publish input images through `image_publisher`, so no camera is required.

<Tabs>
  <Tab title="Image classification">
    Run `sample_resnet101` end-to-end to classify a bundled image with the ResNet-101 model through the Qualcomm® AI Engine Direct SDK (QNN).

    ### Pipeline diagram

    The following diagram shows the ROS 2 nodes and topics in the pipeline.

    ```mermaid theme={null}
    flowchart TD
        A[image_publisher]:::proprietary -->|/image_raw| B[Image classification preprocess]:::proprietary
        B -->|/qrb_inference_input_tensor| C[qrb_ros_nn_inference]:::proprietary
        C -->|/qrb_inference_output_tensor| D[Image classification postprocess]:::proprietary
        D -->|/resnet101_output| E[Consumer or rqt]:::neutral
        classDef proprietary fill:#31017D,color:#ffffff,stroke:#31017D
        classDef neutral fill:#eeeeee,color:#000000,stroke:#999999
    ```

    ### Run the sample

    Follow the OS tab. Both paths launch with the built-in image publisher, which defaults to publishing `glasses.jpg` at 10 Hz.

    <Tabs>
      <Tab title="Qualcomm Linux">
        <Steps>
          <Step title="Source the runtime environment on the device">
            Over SSH to the device:

            ```bash theme={null}
            source /usr/share/qirp-setup.sh -m
            export ROS_DOMAIN_ID=123
            ```
          </Step>

          <Step title="Launch the sample with the image publisher">
            ```bash theme={null}
            ros2 launch sample_resnet101 launch_with_image_publisher.py
            ```
          </Step>

          <Step title="Echo the classification topic">
            In a second SSH session:

            ```bash theme={null}
            ros2 topic echo /resnet101_output
            ```
          </Step>
        </Steps>
      </Tab>

      <Tab title="Ubuntu">
        <Steps>
          <Step title="Install the sample package">
            ```bash theme={null}
            sudo apt install ros-jazzy-sample-resnet101
            ```
          </Step>

          <Step title="Set up the environment on the device">
            ```bash theme={null}
            source /opt/ros/jazzy/setup.bash
            export ROS_DOMAIN_ID=123
            ```
          </Step>

          <Step title="Launch the sample with the image publisher">
            ```bash theme={null}
            ros2 launch sample_resnet101 launch_with_image_publisher.py
            ```
          </Step>

          <Step title="Echo the classification topic">
            ```bash theme={null}
            ros2 topic echo /resnet101_output
            ```
          </Step>
        </Steps>
      </Tab>
    </Tabs>

    ### Expected result

    The following output shows a steady classification stream on `/resnet101_output`.

    ```text theme={null}
    data: 'sunglass'
    ---
    data: 'sunglass'
    ---
    data: 'sunglass'
    ```

    Pass `image_path:=<your image path>` to the launch command to change the input image; the label updates accordingly.
  </Tab>

  <Tab title="Hand detection">
    Run `sample_hand_detection` end-to-end to detect palms and hand landmarks with the MediaPipe Hand Landmark Detector through QNN. The two-stage pipeline runs palm detection followed by landmark regression, then annotates the source frame.

    ### Pipeline diagram

    The following diagram shows the ROS 2 nodes and topics in the two-stage pipeline.

    ```mermaid theme={null}
    flowchart TD
        A[image_publisher]:::proprietary -->|/image_raw| B[palm_preprocessor]:::proprietary
        B -->|/palm_detector_input_tensor| C[qrb_ros_nn_inference]:::proprietary
        C -->|/palm_detector_output_tensor| D[palm_postprocessor]:::proprietary
        D --> E[landmark_preprocessor]:::proprietary
        E -->|/landmark_detector_input_tensor| F[qrb_ros_nn_inference]:::proprietary
        F -->|/landmark_detector_output_tensor| G[landmark_postprocessor]:::proprietary
        G -->|/handlandmark_result| H[rqt Image View]:::neutral
        classDef proprietary fill:#31017D,color:#ffffff,stroke:#31017D
        classDef neutral fill:#eeeeee,color:#000000,stroke:#999999
    ```

    ### Run the sample

    Follow the OS tab. Both paths launch with the built-in image publisher, so no camera is required.

    <Tabs>
      <Tab title="Qualcomm Linux">
        <Steps>
          <Step title="Source the runtime environment on the device">
            Over SSH to the device:

            ```bash theme={null}
            source /usr/share/qirp-setup.sh -m
            export ROS_DOMAIN_ID=123
            export PYTHONPATH=$PYTHONPATH:/opt/ros/jazzy/lib/python3.14/site-packages/sample_hand_detection
            ```
          </Step>

          <Step title="Launch the sample with the image publisher">
            ```bash theme={null}
            ros2 launch sample_hand_detection launch_with_image_publisher.py model_path:=/opt/model/
            ```
          </Step>

          <Step title="View the annotated frames on the host">
            In a host terminal:

            ```bash theme={null}
            export DISPLAY=YOUR_HOST_IP:0
            export ROS_DOMAIN_ID=123
            rqt
            ```

            In `rqt`, select **Plugins** → **Visualization** → **Image View** and choose `/handlandmark_result`.
          </Step>
        </Steps>
      </Tab>

      <Tab title="Ubuntu">
        <Steps>
          <Step title="Install the sample package">
            ```bash theme={null}
            sudo apt install ros-jazzy-sample-hand-detection
            ```
          </Step>

          <Step title="Set up the environment on the device">
            ```bash theme={null}
            source /opt/ros/jazzy/setup.bash
            export ROS_DOMAIN_ID=123
            ```
          </Step>

          <Step title="Launch the sample with the image publisher">
            ```bash theme={null}
            ros2 launch sample_hand_detection launch_with_image_publisher.py
            ```
          </Step>

          <Step title="View the annotated frames on the host">
            On the host, open `rqt`, select **Plugins** → **Visualization** → **Image View**, and choose `/handlandmark_result`.
          </Step>
        </Steps>
      </Tab>
    </Tabs>

    ### Expected result

    The following figure shows the annotated frame with the palm bounding box and the 21 landmark markers when a fully open palm image is published.

    <Frame>
      <img src="https://mintcdn.com/qualcomm-prod/ZRoYdq-twSwPVBFY/snippets/qir-sdk2.0-user-guide/robotics-workflow-unified/media/showcase-hands.png?fit=max&auto=format&n=ZRoYdq-twSwPVBFY&q=85&s=b0cadf2ce74ea94922d027920fbf7adc" style={{width: "50%", display: "block", margin: "0 auto"}} width="208" height="202" data-path="snippets/qir-sdk2.0-user-guide/robotics-workflow-unified/media/showcase-hands.png" />
    </Frame>
  </Tab>
</Tabs>

## Next steps

* [QIR SDK user guide (Qualcomm Linux)](/SDKs/QIR-SDK-2.0/) — Complete QIR SDK user guide for Qualcomm Dragonwing devices running Qualcomm Linux software, including SDK overview, install, image variants, cross-compile toolchain, overlay mechanism, migration guide, and troubleshooting.
* [QIR SDK user guide (Ubuntu)](/SDKs/QIR-SDK-Ubuntu) — Complete QIR SDK user guide for Qualcomm Dragonwing devices running Ubuntu, including software architecture, install, and sample applications.
* Try more sample applications

  <Columns cols={2}>
    <Card title="QIR SDK Sample applications (Linux)" icon="linux">
      [Samples](/SDKs/QIR-SDK-2.0/qir-sdk-sample-applications)
    </Card>

    <Card title="QIR SDK Sample applications (Ubuntu)" icon="ubuntu">
      [Samples](/SDKs/QIR-SDK-Ubuntu/sample-applications)
    </Card>
  </Columns>