CPP yolov8 detection boxes not right position

[MyraBaba]

Hi,

I try to test with my images but boxes detection position all left corner not datected face what is the problem here:
left green

Best

`#include <iostream>
#include <vector>
#include <getopt.h>

#include <opencv2/opencv.hpp>

#include "inference.h"

using namespace std;
using namespace cv;

int main(int argc, char **argv)
{
    std::string projectBasePath = "/home/user/ultralytics"; // Set your ultralytics base path

    bool runOnGPU = true;

    //
    // Pass in either:
    //
    // "yolov8s.onnx" or "yolov5s.onnx"
    //
    // To run Inference with yolov8/yolov5 (ONNX)
    //

    // Note that in this example the classes are hard-coded and 'classes.txt' is a place holder.
    Inference inf("/data/dProjects/yolov8-face-landmarks-opencv-dnn/weights/yolov8n-face.onnx", cv::Size(640, 480), "classes.txt", runOnGPU);

    std::vector<std::string> imageNames;
    imageNames.push_back( "/home/alp2080/face_detector_3.jpg");
    imageNames.push_back( "/home/alp2080/detector_2.jpg");

    for (int i = 0; i < imageNames.size(); ++i)
    {
        cv::Mat frame = cv::imread(imageNames[i]);

        // Inference starts here...
        std::vector<Detection> output = inf.runInference(frame);

        int detections = output.size();
        std::cout << "Number of detections:" << detections << std::endl;

        for (int i = 0; i < detections; ++i)
        {
            Detection detection = output[i];

            cv::Rect box = detection.box;
            cv::Scalar color = detection.color;

            // Detection box
            cv::rectangle(frame, box, cv::Scalar(0, 255, 0), 3);

            // Detection box text
            std::string classString = detection.className + ' ' + std::to_string(detection.confidence).substr(0, 4);
            cv::Size textSize = cv::getTextSize(classString, cv::FONT_HERSHEY_DUPLEX, 1, 2, 0);
            cv::Rect textBox(box.x, box.y - 40, textSize.width + 10, textSize.height + 20);

            cv::rectangle(frame, textBox, color, cv::FILLED);
            cv::putText(frame, classString, cv::Point(box.x + 5, box.y - 10), cv::FONT_HERSHEY_DUPLEX, 1, cv::Scalar(0, 0, 0), 2, 0);
        }
        // Inference ends here...

        // This is only for preview purposes
        float scale = 0.8;
        cv::resize(frame, frame, cv::Size(frame.cols*scale, frame.rows*scale));
        cv::imshow("Inference", frame);

        cv::waitKey(-1);
    }
}
`

[MyraBaba]

@derronqi We are using yolov5face , its detection and landmarks very accurate than the insightface detection but took 20 ms on the Gpu per frame . Slow for us.

We need to utilize 2 rtsp cam and detect faces faster

[Upliner]

I have this bug too

[Upliner]

OK, now I found out that "export" function in this repository is buggy. Install ultralytics from pip to get onnx file.

[MyraBaba]

@Upliner did you compare with scrfd / retina and yolov5face ? quality and speed ?

[Upliner]

@Upliner did you compare with scrfd / retina and yolov5face ? quality and speed ?

I only compared with https://github.com/derronqi/yolov7-face the quality is better and CPU speed is faster, I haven't checked GPU speed yet, I think I'll build GPU version tomorrow.

[Upliner]

@Upliner did you compare with scrfd / retina and yolov5face ? quality and speed ?

I'm using onnxruntime+CUDA on RTX3060 and yolov8n-face model, average inference time is near 9-10ms. Here is 30 time samples in microseconds:
[8338, 13978, 16457, 14625, 8933, 11802, 13985, 6536, 9729, 7278, 7799, 7864, 7995, 7685, 7965, 6473, 16348, 12040, 9911, 6533, 6022, 8061, 6879, 9495, 7139, 6560, 6707, 6536, 7561, 13201]