RZV2L GUI DRP-AI HMI Demo

This project runs the DRP-AI Demos and the DRP-AI Pre-trained Applications. This avoid the need to use a terminal to use a terminal to start the applications. The DRP-AI demos and Applications have been modifed to work with the GUI framework. See below for more information.

Running the applications

Requirements

• Eclipse IDE is needed to Build the project.
• SDL Needs to be installed during Yocto Build.
  • In the Yocto configuration file "build/conf/local.conf"
  • Add IMAGE_INSTALL_append = " libsdl2-dev"
• Install the RZV2L SDK generated from Yocto
• Due to it hub size constraints the following files cannot be uploaded
  • yolov3_cam_weight.dat
  • hrnet_cam_weight.dat

Getting Started

• Step 1) Clone this repository
• Step 2) in a terminal cd to the repository
• Step 3) Setup the SDK i.e. "source /opt/poky/3.1.31/environment-setup-aarch64-poky-linux" NOTE: version 3.1.31 may differ
• Step 4) Start eclipse in the current terminal "eclipse &". NOTE: If you start eclipse outside the terminal in Step 3 SDK will not work.
• Step 5) In the Eclipse Workspace Dialog specify the directory where the repository is located refer to Step 3
• Step 6) Import Project using General -> Existing Projects into Workspace
• Step 7) Right click on each project and select Build.

Deployment

Each of the Applications and their DRP-AI files must be layed out as follows on the RZV2L board as follows.

├── AIDemos
│   ├── Elderly_fall_detection
│   │   ├── 03_Elderly_fall_detection
│   │   ├── hrnet_cam
│   │   └── tinyyolov2_cam
│   ├── Face_recognition
│   │   ├── Face_recognition
│   │   └── resnet50_cam
│   ├── Head_count_cam
│   │   ├── head_count_cam_app
│   │   └── yolov3_cam
│   └── Line_Crossing
│       ├── Line_crossing_object_counting
│       └── tinyyolov2_cam
├── ObjectDetection
│   ├── coco-labels-2014_2017.txt
│   ├── sample_app_tinyyolov2_cam
│   ├── sample_app_tinyyolov3_cam
│   ├── sample_app_yolov2_cam
│   ├── sample_app_yolov3_cam
│   ├── tinyyolov2_cam
│   ├── tinyyolov3_cam
│   ├── yolov2_cam
│   └── yolov3_cam
├── PoseEstimation
│   └── app_hrnet_pre-tinyyolov2_cam
│       ├── hrnet_cam
│       ├── sample_app_hrnet_pre-tinyyolov2_cam
│       └── tinyyolov2_cam
└── rzv2lDemo

RZV2L HMI Application

The HMI Application rzv2lDemo demo was developed with the LVGL GUI Software Framework. This framework rquires SDL installed in the Yocto SDK and BSP.

DRP-AI Applications

All these demos were modified to respond to Mouse or Touch to close the application. The Camera code is modified to support both MIPI and USB Camera.

NOTE: DO NOT CONNECT BOTH CAMERAS TO THE BOARD.

DRP-AI Yolo Demos

The DRP-AI Yolo Application "sample_yolo_cam" project creates demos for YoloV2, YoloV3, TinyYoloV2, and TinyYoloV3 to be used with the Camera.

NOTE: The YoloV5 and YOloVx are still in development.

DRP-AI HRNET

Only Mouse/touch UI funcitonalit and Camera modification was added.

DRP-AI Pre-trained Applications

The DRP-Ai Pre-trained demo's modifications include the mouse/touch response to close the demo and the camera modifcation mentioned in the other demo.