retinaface

RetinaFace Example

Overview

Detect human faces in images and identify face landmark features (eyes, nose, and mouth).

Usage

Make sure you have downloaded the data files first for the examples. You only need to do this once for all examples.

cd example/
git clone --depth=1 https://github.com/swdee/go-rknnlite-data.git data

Run the RetinaFace example on rk3588 or replace with your Platform model.

cd example/retinaface
go run retinaface.go -p rk3588

This will result in the output of:

Driver Version: 0.9.6, API Version: 2.3.0 (c949ad889d@2024-11-07T11:35:33)
Model Input Number: 1, Ouput Number: 3
Input tensors:
  index=0, name=input0, n_dims=4, dims=[1, 320, 320, 3], n_elems=307200, size=307200, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-14, scale=1.074510
Output tensors:
  index=0, name=output0, n_dims=3, dims=[1, 4200, 4, 0], n_elems=16800, size=16800, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=0, scale=0.044699
  index=1, name=572, n_dims=3, dims=[1, 4200, 2, 0], n_elems=8400, size=16800, fmt=UNDEFINED, type=FP16, qnt_type=AFFINE, zp=0, scale=1.000000
  index=2, name=571, n_dims=3, dims=[1, 4200, 10, 0], n_elems=42000, size=42000, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-22, scale=0.086195
Model first run speed: inference=9.732158ms, post processing=163.914µs, rendering=1.654012ms, total time=11.550084ms
face @ (312 531 453 714) 0.998047
face @ (306 289 436 454) 0.997559
face @ (53 286 184 460) 0.996582
face @ (543 533 680 716) 0.996094
face @ (56 537 181 703) 0.995605
face @ (61 34 192 209) 0.994629
face @ (523 274 674 474) 0.994141
face @ (553 28 695 224) 0.991211
face @ (292 36 421 217) 0.991211
Saved object detection result to ../data/face-out.jpg
Benchmark time=573.784348ms, count=100, average total time=5.737843ms
done

The saved JPG image with face landmarks indicated.

See the help for command line parameters.

$ go run retinaface.go --help

Usage of /tmp/go-build3703609935/b001/exe/retinaface:
  -i string
        Image file to run inference on (default "../data/face.jpg")
  -m string
        RKNN compiled Retina Face model file (default "../data/models/rk3588/retinaface-320-rk3588.rknn")
  -o string
        The output JPG file with face detection markers (default "../data/face-out.jpg")
  -p string
        Rockchip CPU Model number [rk3562|rk3566|rk3568|rk3576|rk3582|rk3582|rk3588] (default "rk3588")

Docker

To run the RetinaFace example using the prebuilt docker image, make sure the data files have been downloaded first, then run.

# from project root directory

docker run --rm \
  --device /dev/dri:/dev/dri \
  -v "$(pwd):/go/src/app" \
  -v "$(pwd)/example/data:/go/src/data" \
  -v "/usr/include/rknn_api.h:/usr/include/rknn_api.h" \
  -v "/usr/lib/librknnrt.so:/usr/lib/librknnrt.so" \
  -w /go/src/app \
  swdee/go-rknnlite:latest \
  go run ./example/retinaface/retinaface.go -p rk3588

Benchmarks

The following table shows a comparison of the benchmark results across the three distinct platforms.

Note that these examples are only using a single NPU core to run inference on. The results would be different when running a Pool of models using all NPU cores available. Secondly the Rock 4D (rk3576) has DDR5 memory versus the Rock 5B (rk3588) with slower DDR4 memory.

Background

This RetinaFace example is a Go conversion of the C API example.