Yolov8s.engine文件检测推理结果格式错误

[Yaquarius]

Env

• Ubuntu 20.04
• Cuda 11.4
• TensorRT 8.2.3
  

About this repo

• which model? yolov8s

Your problem

• Hello, when I am testing the detection task with the generated yolov8s.engine file, the reasoning inference result that comes out only shows the prediction box and numeric labels, but not the category and confidence level, what should I do to solve it?

[wang-xinyu]

You need to modify the draw_bbox function

tensorrtx/yolov8/src/postprocess.cpp

Line 120 in 47baebc

void draw_bbox(std::vector<cv::Mat>& img_batch, std::vector<std::vector<Detection>>& res_batch) {