Sample applications for Qualcomm® Linux platforms

Sample applications for Qualcomm® Linux repository provide sample applications to demonstrate the capability of Qualcomm Linux platforms.

Build Sample Applications

Generate the eSDK containing the cross-compiler tool chain

Note: qcom-wayland-x86_64-qcom-multimedia-image-armv8-2a--toolchain-1.0.sh is required to build sample applications.

Follow these build procedures to build the Qualcomm Intelligent Multimedia Product SDK (QIMP SDK).

Run the following command from the build directory to generate the eSDK:

bitbake -c populate_sdk_ext qcom-multimedia-image

qcom-wayland-x86_64-qcom-multimedia-image-armv8-2a--toolchain-1.0.sh is generated at /build-qcom-wayland/tmp-glibc/deploy/sdk.

Extract the toolchain using

./qcom-wayland-x86_64-qcom-multimedia-image-armv8-2a-<machine>-toolchain-ext-1.0.sh

Build Standalone Sample App using makefile.

Hello-QIM is the sample application that can be used to experience the capabilities of the Qualcomm Linux platform. The sample application is hosted on GitHub.

Prerequisites: Before you begin developing the Hello QIM sample application, ensure that you Generate platform SDK on the host machine.

1. Run the following command to go to the directory where the platform SDK was installed. cd

2. Run the following command to set up the source environment: source environment-setup-armv8-2a-qcom-linux

3. git clone https://github.com/quic/sample-apps-for-qualcomm-linux

4. cd sample-apps-for-qualcomm-linux/Hello-QIM

5. export SDKTARGETSYSROOT=/tmp/sysroots

   e.g.: export SDKTARGETSYSROOT=/local/mnt/workspace/Platform_eSDK_plus_QIM/tmp/sysroots

   export MACHINE=

   e.g.: export MACHINE=qcs6490-rb3gen2-vision-kit

   export GST_APP_NAME=

   e.g.: export GST_APP_NAME=gst-appsink

6. make

To run the Hello-QIM program, do the following

8. Run the following command to transfer the program to the Qualcomm Reference kit. adb push gst-appsink /opt/ adb shell chmod 777 /opt/gst-appsink cd /opt/ ./gst-appsink -w 1280 -h 720.

The Hello-QIM application is successfully created. The following message is displayed: Hello-QIM: Success creating pipeline and received camera frame.

Model Details

Contains Qualcomm® Neural Processing SDK quantized models and TensorFlow Lite (TFLite) to execute the Sample applications :

Models are under Qualcomm License. Download the models from the Release

Model Name	SDK	Source Model Path	Source Model License
Object Detection
Yolo-NAS	Neural Processing SDK	Link	License
Classification
InceptionV3	Neural Processing SDK	Link	License
InceptionV3	TensorFlow Lite	Link	License
Image Segmentation
DeeplabV3-resnet50	Neural Processing SDK	Link	License
DeeplabV3-resnet50	TensorFlow Lite	Link	License
Pose Estimation
PoseNet Mobilenet V1	TensorFlow Lite	link	License

License

This project is license under the BSD-3-Clause-Clear license. See LICENSE for the full license text.