好文转载：基于深度学习的水果检测与识别系统（Python界面版，YOLOv5实现）_from utils.torch_utils import time_synchronized

import argparse
import random
1
2

import cv2
import numpy as np
import torch

from models.experimental import attempt_load
from utils.datasets import letterbox
from utils.general import non_max_suppression, check_img_size, scale_coords
from utils.torch_utils import time_synchronized, select_device

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11

        然后，我们需要定义一些函数，其中predict()函数用于实现图片的预测推理；cv_imread()函数用于读取图片；plot_one_box()函数用于绘制预测框。

def predict(img):
    img = torch.from_numpy(img).to(device)
    img = img.half() if half else img.float()
    img /= 255.0
    if img.ndimension() == 3:
        img = img.unsqueeze(0)
t1 <span class="token operator">=</span> time_synchronized<span class="token punctuation">(</span><span class="token punctuation">)</span>
pred <span class="token operator">=</span> model<span class="token punctuation">(</span>img<span class="token punctuation">,</span> augment<span class="token operator">=</span><span class="token boolean">False</span><span class="token punctuation">)</span><span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span>
pred <span class="token operator">=</span> non_max_suppression<span class="token punctuation">(</span>pred<span class="token punctuation">,</span> opt<span class="token punctuation">.</span>conf_thres<span class="token punctuation">,</span> opt<span class="token punctuation">.</span>iou_thres<span class="token punctuation">,</span> classes<span class="token operator">=</span>opt<span class="token punctuation">.</span>classes<span class="token punctuation">,</span>
                           agnostic<span class="token operator">=</span>opt<span class="token punctuation">.</span>agnostic_nms<span class="token punctuation">)</span>
t2 <span class="token operator">=</span> time_synchronized<span class="token punctuation">(</span><span class="token punctuation">)</span>
InferNms <span class="token operator">=</span> <span class="token builtin">round</span><span class="token punctuation">(</span><span class="token punctuation">(</span>t2 <span class="token operator">-</span> t1<span class="token punctuation">)</span><span class="token punctuation">,</span> <span class="token number">2</span><span class="token punctuation">)</span>

<span class="token keyword">return</span> pred<span class="token punctuation">,</span> InferNms
1
2
3
4
5
6
7
8
1
2
3
4
5
6

def cv_imread(filePath):
# 读取图片
cv_img = cv2.imdecode(np.fromfile(filePath, dtype=np.uint8), -1)

<span class="token keyword">if</span> <span class="token builtin">len</span><span class="token punctuation">(</span>cv_img<span class="token punctuation">.</span>shape<span class="token punctuation">)</span> <span class="token operator">&gt;</span> <span class="token number">2</span><span class="token punctuation">:</span>
    <span class="token keyword">if</span> cv_img<span class="token punctuation">.</span>shape<span class="token punctuation">[</span><span class="token number">2</span><span class="token punctuation">]</span> <span class="token operator">&gt;</span> <span class="token number">3</span><span class="token punctuation">:</span>
        cv_img <span class="token operator">=</span> cv_img<span class="token punctuation">[</span><span class="token punctuation">:</span><span class="token punctuation">,</span> <span class="token punctuation">:</span><span class="token punctuation">,</span> <span class="token punctuation">:</span><span class="token number">3</span><span class="token punctuation">]</span>
<span class="token keyword">return</span> cv_img
1
2
3
4

def plot_one_box(img, x, color=None, label=None, line_thickness=None):
# Plots one bounding box on image img
tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1 # line/font thickness
color = color or [random.randint(0, 255) for _ in range(3)]
c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA)
if label:
tf = max(tl - 1, 1) # font thickness
t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA) # filled
cv2.putText(img, label, (c1[0], c1[1] - 2), 0, tl / 3, [225, 255, 255], thickness=tf, lineType=cv2.LINE_AA)

if name == ‘main’:
img_path = “./UI_rec/test_/b-1459-_jpg.rf.06e32fd1fbee5a10c776e1e63237904f.jpg”
image = cv_imread(img_path)
image = cv2.resize(image, (850, 500))
img0 = image.copy()
img = letterbox(img0, new_shape=imgsz)[0]
img = np.stack(img, 0)
img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB, to 3x416x416
img = np.ascontiguousarray(img)

pred<span class="token punctuation">,</span> useTime <span class="token operator">=</span> predict<span class="token punctuation">(</span>img<span class="token punctuation">)</span>

det <span class="token operator">=</span> pred<span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span>
p<span class="token punctuation">,</span> s<span class="token punctuation">,</span> im0 <span class="token operator">=</span> <span class="token boolean">None</span><span class="token punctuation">,</span> <span class="token string">''</span><span class="token punctuation">,</span> img0
<span class="token keyword">if</span> det <span class="token keyword">is</span> <span class="token keyword">not</span> <span class="token boolean">None</span> <span class="token keyword">and</span> <span class="token builtin">len</span><span class="token punctuation">(</span>det<span class="token punctuation">)</span><span class="token punctuation">:</span>  <span class="token comment"># 如果有检测信息则进入</span>
    det<span class="token punctuation">[</span><span class="token punctuation">:</span><span class="token punctuation">,</span> <span class="token punctuation">:</span><span class="token number">4</span><span class="token punctuation">]</span> <span class="token operator">=</span> scale_coords<span class="token punctuation">(</span>img<span class="token punctuation">.</span>shape<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">:</span><span class="token punctuation">]</span><span class="token punctuation">,</span> det<span class="token punctuation">[</span><span class="token punctuation">:</span><span class="token punctuation">,</span> <span class="token punctuation">:</span><span class="token number">4</span><span class="token punctuation">]</span><span class="token punctuation">,</span> im0<span class="token punctuation">.</span>shape<span class="token punctuation">)</span><span class="token punctuation">.</span><span class="token builtin">round</span><span class="token punctuation">(</span><span class="token punctuation">)</span>  <span class="token comment"># 把图像缩放至im0的尺寸</span>
    number_i <span class="token operator">=</span> <span class="token number">0</span>  <span class="token comment"># 类别预编号</span>
    detInfo <span class="token operator">=</span> <span class="token punctuation">[</span><span class="token punctuation">]</span>
    <span class="token keyword">for</span> <span class="token operator">*</span>xyxy<span class="token punctuation">,</span> conf<span class="token punctuation">,</span> cls <span class="token keyword">in</span> <span class="token builtin">reversed</span><span class="token punctuation">(</span>det<span class="token punctuation">)</span><span class="token punctuation">:</span>  <span class="token comment"># 遍历检测信息</span>
        c1<span class="token punctuation">,</span> c2 <span class="token operator">=</span> <span class="token punctuation">(</span><span class="token builtin">int</span><span class="token punctuation">(</span>xyxy<span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">,</span> <span class="token builtin">int</span><span class="token punctuation">(</span>xyxy<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">)</span><span class="token punctuation">,</span> <span class="token punctuation">(</span><span class="token builtin">int</span><span class="token punctuation">(</span>xyxy<span class="token punctuation">[</span><span class="token number">2</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">,</span> <span class="token builtin">int</span><span class="token punctuation">(</span>xyxy<span class="token punctuation">[</span><span class="token number">3</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">)</span>
        <span class="token comment"># 将检测信息添加到字典中</span>
        detInfo<span class="token punctuation">.</span>append<span class="token punctuation">(</span><span class="token punctuation">[</span>names<span class="token punctuation">[</span><span class="token builtin">int</span><span class="token punctuation">(</span>cls<span class="token punctuation">)</span><span class="token punctuation">]</span><span class="token punctuation">,</span> <span class="token punctuation">[</span>c1<span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span><span class="token punctuation">,</span> c1<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">]</span><span class="token punctuation">,</span> c2<span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span><span class="token punctuation">,</span> c2<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">]</span><span class="token punctuation">]</span><span class="token punctuation">,</span> <span class="token string">'%.2f'</span> <span class="token operator">%</span> conf<span class="token punctuation">]</span><span class="token punctuation">)</span>
        number_i <span class="token operator">+=</span> <span class="token number">1</span>  <span class="token comment"># 编号数+1</span>

        label <span class="token operator">=</span> <span class="token string">'%s %.2f'</span> <span class="token operator">%</span> <span class="token punctuation">(</span>names<span class="token punctuation">[</span><span class="token builtin">int</span><span class="token punctuation">(</span>cls<span class="token punctuation">)</span><span class="token punctuation">]</span><span class="token punctuation">,</span> conf<span class="token punctuation">)</span>

        <span class="token comment"># 画出检测到的目标物</span>
        plot_one_box<span class="token punctuation">(</span>image<span class="token punctuation">,</span> xyxy<span class="token punctuation">,</span> label<span class="token operator">=</span>label<span class="token punctuation">,</span> color<span class="token operator">=</span>colors<span class="token punctuation">[</span><span class="token builtin">int</span><span class="token punctuation">(</span>cls<span class="token punctuation">)</span><span class="token punctuation">]</span><span class="token punctuation">)</span>
<span class="token comment"># 实时显示检测画面</span>
cv2<span class="token punctuation">.</span>imshow<span class="token punctuation">(</span><span class="token string">'Stream'</span><span class="token punctuation">,</span> image<span class="token punctuation">)</span>
<span class="token comment"># if cv2.waitKey(1) &amp; 0xFF == ord('q'):</span>
<span class="token comment">#     break</span>
c <span class="token operator">=</span> cv2<span class="token punctuation">.</span>waitKey<span class="token punctuation">(</span><span class="token number">0</span><span class="token punctuation">)</span> <span class="token operator">&amp;</span> <span class="token number">0xff</span>
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46
• 47
• 48
• 49
• 50
• 51
• 52
• 53
• 54
• 55
• 56
• 57
• 58
• 59
• 60
• 61
• 62
• 63
• 64
• 65
• 66
• 67
• 68
• 69
• 70
• 71
• 72
• 73
• 74

        在这段Python代码中，我们首先定义了一个名为predict的函数，用于对输入的图像进行检测和分类。该函数输入一个图像，先将其转化为PyTorch Tensor，然后对图像进行预测和非极大值抑制，得到预测结果和推理时间。接下来我们定义了cv_imread函数，用于读取图片。cv_imread函数通过OpenCV库的imdecode函数读取图片，并将其转化为NumPy数组，然后返回该数组。接着我们定义了plot_one_box函数，该函数用于在图像上画出检测框和类别标签，其中函数参数中包括了待画出的图像、检测框坐标、类别标签、颜色等。

        最后，在if name == ‘main’:语句中，我们首先读取一张测试图片，并将其调整为合适的大小，然后将其输入到predict函数中进行检测和分类，得到检测结果和推理时间。接着，我们通过遍历检测结果，将检测到的目标物的信息添加到detInfo列表中，然后调用plot_one_box函数，将目标物的检测框和类别标签画在图像上。最后通过OpenCV库的imshow函数将结果图像实时显示在屏幕上。

        这部分代码是我们整个系统实现的关键部分，通过修改和集成该部分代码同样可以实现对输入视频和实时摄像头画面的实时检测和分类，并且可以通过界面将结果显示给用户，提高了系统的实用性。

5.2 PyQt界面设计

        PyQt是Python语言的GUI编程解决方案之一，可以快速地为Python程序创建GUI应用。在本文中，我们使用PyQt库创建一个图形化界面，为用户提供简单易用的交互界面，实现用户选择图片、视频或实时视频进行目标检测。这里介绍系统的GUI设计和开发过程，包括主窗口的设计、UI控件的布局以及信号槽的连接。

        在本系统中，使用PyQt5库设计了可视化的GUI界面，主要包括以下六个模块：

（1）图片检测模块：用户可以选择本地的一张图片进行检测，检测结果将在界面中实时显示。用户也可以选择多张图片进行批量检测，并将结果记录保存到本地。
（2）视频检测模块：用户可以选择本地的一个视频进行检测，检测结果将在界面中实时显示。用户也可以选择多个视频进行批量检测，并将结果记录保存到本地。
（3）实时检测模块：用户可以启动摄像头进行实时检测，检测结果将在界面中实时显示。
（4）更换模型模块：用户可以选择不同的预训练模型进行检测，系统提供了多种不同的预训练模型供用户选择。
（5）结果记录回看模块：用户可以查看历史检测结果，并对检测结果进行搜索和筛选。
（6）其他模块：界面中还包括了一些辅助模块，如登录注册、设置等。

        我们使用Qt Designer设计图形界面，然后使用PyQt将设计好的UI文件转换为Python代码。图形界面中包含多个UI控件，例如：菜单栏、工具栏、标签、按钮等。通过PyQt中的信号槽机制，可以使得UI控件与程序逻辑代码相互连接。

        界面设计使用了Qt Designer工具，通过拖拽组件的方式进行布局，同时也可以通过代码进行动态添加和修改。在界面的设计过程中，我们主要使用了以下几个组件：

（1）QLabel：用于显示文字和图片。
（2）QToolButton：用于实现按钮功能。
（3）QComboBox：用于实现下拉框功能，供用户选择预训练模型。
（4）QFileDialog：用于选择本地的图片和视频。
（5）QTableWidget：用于展示历史检测结果，并提供搜索和筛选功能。

        为了使得界面更加美观和易于操作，我们还可以自定义控件样式，例如在系统中使用的按钮等，都进行了样式的调整。具体实现方式是通过QSS文件，使用CSS语法进行样式定义。

        在图形界面的实现过程中，我们还需要使用Python中的OpenCV库实现视频的读取和显示。使用OpenCV可以快速实现视频的读取、播放和保存等操作，并且可以与PyQt进行结合，实现更为高效的图像和视频处理。下面我们给出系统GUI的示例图：

6. 实验结果与分析

        在实验结果与分析部分，我们不仅通过精度和召回率等指标来评估模型的性能，还通过损失曲线和PR曲线来分析训练过程

        我们使用了前面介绍的水果检测数据集进行训练，该数据集包含多种不同的水果类别，包括苹果、香蕉、橙子、西瓜等，共计8个类别。我们使用了YOLOv5算法，训练了300个epochs，训练过程截图如下。

        在训练过程中，我们使用tensorboard记录了模型在训练集和验证集上的损失曲线。从图6-1中可以看出，随着训练次数的增加，模型的训练损失和验证损失都逐渐降低，说明模型不断地学习到更加精准的特征。

        在训练结束后，我们对模型在测试集上进行了评估，得到了以下结果。下图展示了我们训练的YOLOv5模型在测试集上的PR曲线。可以看到，模型在不同类别上都取得了较高的召回率和精确率，整体表现良好。

        综上，我们训练的YOLOv5模型在水果检测数据集上表现良好，具有较高的检测精度和鲁棒性，可以在实际场景中应用。

        博主对整个系统进行了详细测试，最终开发出一版流畅得到清新界面，就是博文演示部分的展示，完整的UI界面、测试图片视频、代码文件，以及Python离线依赖包（方便安装运行，也可自行配置环境），均已打包上传，感兴趣的朋友可以通过下载链接获取。

下载链接

    若您想获得博文中涉及的实现完整全部程序文件（包括测试图片、视频，py, UI文件等，如下图），这里已打包上传至博主的面包多平台，见可参考博客与视频，已将所有涉及的文件同时打包到里面，点击即可运行，完整文件截图如下：

    在文件夹下的资源显示如下，下面的链接中也给出了Python的离线依赖包，读者可在正确安装Anaconda和Pycharm软件后，复制离线依赖包至项目目录下进行安装，离线依赖的使用详细演示也可见本人B站视频：win11从头安装软件和配置环境运行深度学习项目、Win10中使用pycharm和anaconda进行python环境配置教程。

注意：该代码采用Pycharm+Python3.8开发，经过测试能成功运行，运行界面的主程序为runMain.py和LoginUI.py，测试图片脚本可运行testPicture.py，测试视频脚本可运行testVideo.py。为确保程序顺利运行，请按照requirements.txt配置Python依赖包的版本。Python版本：3.8，请勿使用其他版本，详见requirements.txt文件；

完整资源中包含数据集及训练代码，环境配置与界面中文字、图片、logo等的修改方法请见视频，项目完整文件下载请见参考博客文章里面，或参考视频的简介处给出：➷➷➷

参考博客文章：https://zhuanlan.zhihu.com/p/626198529

参考视频演示：https://www.bilibili.com/video/BV1rT411h7dY/

离线依赖库下载链接：https://pan.baidu.com/s/1hW9z9ofV1FRSezTSj59JSg?pwd=oy4n （提取码：oy4n ）

界面中文字、图标和背景图修改方法：

        在Qt Designer中可以彻底修改界面的各个控件及设置，然后将ui文件转换为py文件即可调用和显示界面。如果只需要修改界面中的文字、图标和背景图的，可以直接在ConfigUI.config文件中修改，步骤如下：
        （1）打开UI_rec/tools/ConfigUI.config文件，若乱码请选择GBK编码打开。
        （2）如需修改界面文字，只要选中要改的字符替换成自己的就好。
        （3）如需修改背景、图标等，只需修改图片的路径。例如，原文件中的背景图设置如下：

mainWindow = :/images/icons/back-image.png

 
 

 
 
1
1

        可修改为自己的名为background2.png图片（位置在UI_rec/icons/文件夹中），可将该项设置如下即可修改背景图：

mainWindow = ./icons/background2.png

 
 

 
 
1
1

结束语

        由于博主能力有限，博文中提及的方法即使经过试验，也难免会有疏漏之处。希望您能热心指出其中的错误，以便下次修改时能以一个更完美更严谨的样子，呈现在大家面前。同时如果有更好的实现方法也请您不吝赐教。

参考文献

[1] Bargoti, S., & Underwood, J. P. (2017). Image segmentation for fruit detection and yield estimation in apple orchards. Journal of Field Robotics, 34(6), 1039-1060.
[2] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587).
[3] Girshick, R. (2015). Fast R-CNN. In Proceedings of the IEEE international conference on computer vision (pp. 1440-1448).
[4] Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster R-CNN: Towards real-time object detection with region proposal networks. In Advances in neural information processing systems (pp. 91-99).
[5] Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C. Y., & Berg, A. C. (2016, October). SSD: Single shot multibox detector. In European conference on computer vision (pp. 21-37). Springer, Cham.
[6] Lin, T. Y., Goyal, P., Girshick, R., He, K., & Dollár, P. (2017). Focal loss for dense object detection. In Proceedings of the IEEE international conference on computer vision (pp. 2980-2988).
[7] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
[8] Redmon, J., & Farhadi, A. (2017). YOLO9000: Better, faster, stronger. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7263-7271).
[9] Redmon, J., & Farhadi, A. (2018). YOLOv3: An incremental improvement. arXiv preprint arXiv:1804.02767.
[10] Bochkovskiy, A., Wang, C. Y., & Liao, H. Y. M. (2020). YOLOv4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934.
[11] Sa, I., Ge, Z., Dayoub, F., Upcroft, B., Perez, T., & McCool, C. (2016). DeepFruits: A fruit detection system using deep neural networks. Sensors, 16(8), 1222.
[12] Rahnemoonfar, M., & Sheppard, C. (2017). Deep count: Fruit counting based on deep simulated learning. Sensors, 17(4), 905.
[13] Chakraborty S K, Subeesh A, Dubey K, et al. Development of an optimally designed real-time automatic citrus fruit grading–sorting​ machine leveraging computer vision-based adaptive deep learning model[J]. Engineering Applications of Artificial Intelligence, 2023, 120: 105826.
[14] Jocher, G., Stoken, A., Borovec, J., Bendahan, R. A., Puig, D., Tokozume, Y., … & Guan, H. (2021). Ultralytics/yolov5: v5.0-YOLOv5-PyTorch.