Skip to content

README.md

Latest commit

 

History

History
92 lines (62 loc) · 2.31 KB

README.md

File metadata and controls

92 lines (62 loc) · 2.31 KB

RZ/G3E Face Detection Application

Overview

This application performs face detection on RZ/G3E using either a USB camera stream or a still image file. Detection results are shown on an HDMI display.

The model is compiled with the RUHMI AI Framework and executed with Ethos-U55 acceleration.

Operation flow

Face Detection Flow

Target Environment

  • Board: RZ/G3E-EVKIT
  • Software: RZ/G3E Ethos-U Package (including RUHMI runtime)
  • Peripherals:
    • USB camera
    • HDMI display
    • microSD card (optional)

System configuration:

Face Detection System

Directory Structure

.
├── README.md                           // This document
├── exe
│   ├── face_detection                  // Application binary
│   └── model_yolo-fastest              // AI model directory, compiled using the RUHMI AI Framework
│          ├── config.yaml
│          └── yolo-fastest_192_face_v4
└── src	                                // Application source code

exe/ and src/ are not included in this repository. Use the RZ/G3E release package for runnable assets.

Model Information

AI Model Input size Output size
yolo-fastest_192_face_v4.tflite int8 [1,192,192,1] int8[1,6,6,18]
int8[1,12,12,18]

Build

Build is required only when src/ is included in your release package.

  1. Install and source the RZ/G3E toolchain environment.
  2. Build the application:
mkdir -p src/build
cd src/build
cmake ..
make

Generated binary: src/build/face_detection

Run

Copy files to RZ/G3E-EVKIT:

scp -r exe/ root@<TARGET_IP>:/home/root/

USB camera mode:

./face_detection USB

Image file mode:

./face_detection IMAGE <path_to_image>

Expected output includes model info, FPS, and number of faces detected.

Notes

  • FPS values are reference values only.
  • Press Enter in the running console to terminate the app.
  • Refer to LICENSE in the repository root for license information.