go-rknnlite/example/yolov8

YOLOv8 Example

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 YOLOv8 example on rk3588 or replace with your Platform model.

cd example/yolov8
go run yolov8.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: 9
Input tensors:
  index=0, name=images, n_dims=4, dims=[1, 640, 640, 3], n_elems=1228800, size=1228800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003922
Output tensors:
  index=0, name=318, n_dims=4, dims=[1, 64, 80, 80], n_elems=409600, size=409600, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-56, scale=0.110522
  index=1, name=onnx::ReduceSum_326, n_dims=4, dims=[1, 80, 80, 80], n_elems=512000, size=512000, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003452
  index=2, name=331, n_dims=4, dims=[1, 1, 80, 80], n_elems=6400, size=6400, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003482
  index=3, name=338, n_dims=4, dims=[1, 64, 40, 40], n_elems=102400, size=102400, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-17, scale=0.098049
  index=4, name=onnx::ReduceSum_346, n_dims=4, dims=[1, 80, 40, 40], n_elems=128000, size=128000, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003592
  index=5, name=350, n_dims=4, dims=[1, 1, 40, 40], n_elems=1600, size=1600, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003755
  index=6, name=357, n_dims=4, dims=[1, 64, 20, 20], n_elems=25600, size=25600, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-49, scale=0.078837
  index=7, name=onnx::ReduceSum_365, n_dims=4, dims=[1, 80, 20, 20], n_elems=32000, size=32000, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003817
  index=8, name=369, n_dims=4, dims=[1, 1, 20, 20], n_elems=400, size=400, fmt=NCHW, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003835
person @ (109 237 225 535) 0.896928
bus @ (92 136 548 439) 0.885478
person @ (475 233 559 521) 0.881662
person @ (211 241 285 510) 0.832044
person @ (80 326 123 517) 0.596252
Model first run speed: inference=41.492251ms, post processing=1.968418ms, rendering=714.861µs, total time=44.17553ms
Saved object detection result to ../data/bus-yolov8-out.jpg
Benchmark time=3.926217098s, count=100, average total time=39.26217ms
done

The saved JPG image with object detection markers.

To use your own RKNN compiled model and images.

go run yolov8.go -m <RKNN model file> -i <image file> -l <labels txt file> -o <output jpg file> -p <platform>

The labels file should be a text file containing the labels the Model was trained on. It should have one label per line.

See the help for command line parameters.

$ go run yolov8.go --help

Usage of /tmp/go-build156306685/b001/exe/yolov8:
  -i string
        Image file to run object detection on (default "../data/bus.jpg")
  -l string
        Text file containing model labels (default "../data/coco_80_labels_list.txt")
  -m string
        RKNN compiled YOLO model file (default "../data/models/rk3588/yolov8s-rk3588.rknn")
  -o string
        The output JPG file with object detection markers (default "../data/bus-yolov8-out.jpg")
  -p string
        Rockchip CPU Model number [rk3562|rk3566|rk3568|rk3576|rk3582|rk3582|rk3588] (default "rk3588")

Docker

To run the YOLOv8 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/yolov8/yolov8.go -p rk3588

Proprietary Models

The example YOLOv8 model used has been trained on the COCO dataset so makes use of the default Post Processor setup. If you have trained your own Model and have set specific Classes or want to use alternative Box and NMS Threshold values, then initialize the postprocess.NewYOLOv8 with your own YOLOv8Params.

In the file postprocess/yolov8.go see function YOLOv8COCOParams for how to configure your own custom parameters.

Benchmarks

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

Platform	Execution Time	Average Inference Time Per Image
rk3588	3.92s	39.26ms
rk3576	3.94s	39.48ms
rk3566	8.86s	88.66ms

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 YOLOv8 example is a Go conversion of the C API example.