ios部署yolov5

最新推荐文章于 2024-08-08 08:33:46 发布

独孤的大山猫

最新推荐文章于 2024-08-08 08:33:46 发布

阅读量2.7k

点赞数 3

分类专栏： Useful_tools&skills 文章标签： ios 部署yolov5 yolov5.h 计算机视觉机器学习

本文链接：https://blog.csdn.net/xiqi4145/article/details/115273246

版权

Useful_tools&skills 专栏收录该内容

82 篇文章 6 订阅

订阅专栏

本文档详细介绍了将YOLOv5模型转换为ONNX格式，并使用NCNN进行移动端部署的步骤。首先，通过检测脚本验证权重文件的正确性，然后利用export.py将pt模型转换为onnx，再用monnxsim简化模型。在转换过程中可能会遇到不支持的slice step问题。最后，展示了YoloV5类的定义，包括模型结构、类标签和输出层信息，适用于iOS平台的物体检测应用。

摘要由CSDN通过智能技术生成

参考网址：https://github.com/cmdbug/YOLOv5_NCNN

这里这三个输出层的名称，每一个人转出来的不一定一样。

结构查看： https://netron.app/

查看best.pt转换生成的param文件结构，然后修改

标签名，类名，输出层

//
//  YoloV5.hpp
//  YOLOv5NCNN
//
//  Created by WZTENG on 2020/7/6.
//  Copyright © 2020 TENG. All rights reserved.
//

#ifndef YoloV5_hpp
#define YoloV5_hpp

#include <stdio.h>
#include "ncnn/ncnn/net.h"
#import <Foundation/Foundation.h>
#import <UIKit/UIKit.h>
#import <UIKit/UIImage.h>
#import <functional>

namespace yolocv {
    typedef struct {
        int width;
        int height;
    }YoloSize;
}

typedef struct {
    std::string name;
    int stride;
    std::vector<yolocv::YoloSize> anchors;
}YoloLayerData;

typedef struct BoxInfo {
    float x1;
    float y1;
    float x2;
    float y2;
    float score;
    int label;
}BoxInfo;

class YoloV5 {
public:
    YoloV5(bool useGPU);
    ~YoloV5();
    std::vector<BoxInfo> dectect(UIImage *image, float threshold, float nms_threshold);
    //改成自己的类别
    std::vector<std::string> labels{"person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light",
        "fire hydrant", "stop sign", "parking meter", "bench", "bird", "cat", "dog", "horse", "sheep", "cow",
        "elephant", "bear", "zebra", "giraffe", "backpack", "umbrella", "handbag", "tie", "suitcase", "frisbee",
        "skis", "snowboard", "sports ball", "kite", "baseball bat", "baseball glove", "skateboard", "surfboard",
        "tennis racket", "bottle", "wine glass", "cup", "fork", "knife", "spoon", "bowl", "banana", "apple",
        "sandwich", "orange", "broccoli", "carrot", "hot dog", "pizza", "donut", "cake", "chair", "couch",
        "potted plant", "bed", "dining table", "toilet", "tv", "laptop", "mouse", "remote", "keyboard", "cell phone",
        "microwave", "oven", "toaster", "sink", "refrigerator", "book", "clock", "vase", "scissors", "teddy bear",
        "hair drier", "toothbrush"};
    
//    std::vector<std::string> labels{"sender", "receiver"};

private:
    static std::vector<BoxInfo> decode_infer(ncnn::Mat &data, int stride, const yolocv::YoloSize& frame_size, int net_size, int num_classes, const std::vector<yolocv::YoloSize>& anchors, float threshold);
    static void nms(std::vector<BoxInfo>& result, float nms_threshold);
    
    ncnn::Net* Net;
    int input_size = 640;
    int num_class = 80;  // 修改成自己的类别
    std::vector<YoloLayerData> layers{  // 根据param文件去改
        {"394", 32, {{116, 90}, {156, 198}, {373, 326}}},
        {"375", 16, {{30, 61}, {62, 45}, {59, 119}}},
        {"output", 8, {{10, 13}, {16, 30}, {33, 23}}},
    };
    
public:
    static YoloV5 *detector;
    static bool hasGPU;
    static bool toUseGPU;
    
};


#endif /* YoloV5_hpp */