[Solved] C++ reason ncnn model error: Segmentation fault (core dumped)

The reasoning code is as follows:

#include "net.h"
#include <iostream>
#include <fstream>
#include <algorithm>
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgproc.hpp>
#include <stdio.h>
#include <vector>
#include <opencv2/opencv.hpp>

using namespace std;

void pretty_print(const ncnn::Mat& m)
    for (int q=0; q<m.c; q++)
        const float* ptr = m.channel(q);
        for (int y=0; y<m.h; y++)
            for (int x=0; x<m.w; x++)
                printf("%f ", ptr[x]);
            ptr += m.w;
//main fuction
int main(){
    string img_path = "person1.jpeg";
    cv::Mat img = cv::imread(img_path, cv::IMREAD_COLOR);
    cv::Mat img2;
    int input_width = 512;//Input size specified when going onnx
    int input_height = 512;
    cv::resize(img, img2, cv::Size(input_width, input_height));

    // Load the converted and quantized alexnet network
    ncnn::Net net;
    // net.opt.num_threads=1;
    // Convert opencv mat to ncnn mat
    ncnn::Mat input = ncnn::Mat::from_pixels(img2.data, ncnn::Mat::PIXEL_BGR, img2.cols, img2.rows);
    const float mean_vals[3] = {0.485, 0.456, 0.406};
    const float norm_vals[3] = {0.229, 0.224, 0.225}; //[0.485, 0.456, 0.406]),std=np.array([0.229, 0.224, 0.225]
    input.substract_mean_normalize(mean_vals, norm_vals);
    // ncnn forward calculation
    ncnn::Extractor extractor = net.create_extractor();
    extractor.input("input.1", input);
    ncnn::Mat output0;
    extractor.extract("1035", output0);

    //ncnn::mat ->>>>> cv::mat
    cv::Mat a(input_height,input_width, CV_8UC3);
    output0.to_pixels(a.data, ncnn::Mat::PIXEL_BGR2RGB);
    cv::imwrite("ncnninfer.png", a);
    // pretty_print(output0);
    // pretty_print(output1);

    return 0;

Only segmentation fault (core dumped) is reported after running

No core file is generated. Check through ulimit -c, the size of core file is unlimited and there is no problem. This method solves the problem of not generating core file

With core file

Pass under the terminal

apt-get update
apt-get install gdb

Install GDB

Installation completed, passed

gdb ./Execution file name   core

The core here can be modified according to its own core file name.
the error is found at:

Corresponding to the reasoning code above

output0.to_pixels(a.data, ncnn::Mat::PIXEL_BGR2RGB);

It is the code to realize the conversion from ncnn:: mat to cv:mat. After debugging, it is found that the reasoning results of ncnn are all Nan, which leads to the conversion failure

Adjust the previous input, normalization is not done well.

Read More: