新手学习yolov8目标检测小记1

一、复现

网上有很多教程，复现yolov8的目标检测。在复现的过程中，会用到模型yolov8n.pt，可以选择命令下载和网站下载。复现后，runs文件包下会生成最优的权重文件best.py，在ultralytics/assets中放一张图片，按照remead指示，输入命令，用权重文件best.py做预测。运行之后，runs/detect/predict中的图片就是预测识别之后的。

二、数据集

1、准备

搜集一些你需要进行目标检测的图片，比如网上下载、自行拍摄等，可以使用旋转个、放大、缩小、调亮度的方式，对数据进行扩增。代码如下：需要修改第83行的文件夹路径，改为自己存放图片的路径即可。

# -*- coding: utf-8 -*-import cv2
import numpy as np
import os.path
import copy# 椒盐噪声
def SaltAndPepper(src, percetage):SP_NoiseImg = src.copy()SP_NoiseNum = int(percetage * src.shape[0] * src.shape[1])for i in range(SP_NoiseNum):randR = np.random.randint(0, src.shape[0] - 1)randG = np.random.randint(0, src.shape[1] - 1)randB = np.random.randint(0, 3)if np.random.randint(0, 1) == 0:SP_NoiseImg[randR, randG, randB] = 0else:SP_NoiseImg[randR, randG, randB] = 255return SP_NoiseImg# 高斯噪声
def addGaussianNoise(image, percetage):G_Noiseimg = image.copy()w = image.shape[1]h = image.shape[0]G_NoiseNum = int(percetage * image.shape[0] * image.shape[1])for i in range(G_NoiseNum):temp_x = np.random.randint(0, h)temp_y = np.random.randint(0, w)G_Noiseimg[temp_x][temp_y][np.random.randint(3)] = np.random.randn(1)[0]return G_Noiseimg# 昏暗
def darker(image, percetage=0.9):image_copy = image.copy()w = image.shape[1]h = image.shape[0]# get darkerfor xi in range(0, w):for xj in range(0, h):image_copy[xj, xi, 0] = int(image[xj, xi, 0] * percetage)image_copy[xj, xi, 1] = int(image[xj, xi, 1] * percetage)image_copy[xj, xi, 2] = int(image[xj, xi, 2] * percetage)return image_copy# 亮度
def brighter(image, percetage=1.5):image_copy = image.copy()w = image.shape[1]h = image.shape[0]# get brighterfor xi in range(0, w):for xj in range(0, h):image_copy[xj, xi, 0] = np.clip(int(image[xj, xi, 0] * percetage), a_max=255, a_min=0)image_copy[xj, xi, 1] = np.clip(int(image[xj, xi, 1] * percetage), a_max=255, a_min=0)image_copy[xj, xi, 2] = np.clip(int(image[xj, xi, 2] * percetage), a_max=255, a_min=0)return image_copy# 旋转
def rotate(image, angle, center=None, scale=1.0):(h, w) = image.shape[:2]# If no rotation center is specified, the center of the image is set as the rotation centerif center is None:center = (w / 2, h / 2)m = cv2.getRotationMatrix2D(center, angle, scale)rotated = cv2.warpAffine(image, m, (w, h))return rotated# 翻转
def flip(image):flipped_image = np.fliplr(image)return flipped_image# 图片文件夹路径
file_dir = r'E:\yolo_data\mx/'
for img_name in os.listdir(file_dir):img_path = file_dir + img_nameimg = cv2.imread(img_path)# cv2.imshow("1",img)# cv2.waitKey(5000)# 旋转rotated_90 = rotate(img, 90)cv2.imwrite(file_dir + img_name[0:-4] + '_r90.jpg', rotated_90)rotated_180 = rotate(img, 180)cv2.imwrite(file_dir + img_name[0:-4] + '_r180.jpg', rotated_180)for img_name in os.listdir(file_dir):img_path = file_dir + img_nameimg = cv2.imread(img_path)# 镜像flipped_img = flip(img)cv2.imwrite(file_dir + img_name[0:-4] + '_fli.jpg', flipped_img)# 增加噪声# img_salt = SaltAndPepper(img, 0.3)# cv2.imwrite(file_dir + img_name[0:7] + '_salt.jpg', img_salt)img_gauss = addGaussianNoise(img, 0.3)cv2.imwrite(file_dir + img_name[0:-4] + '_noise.jpg', img_gauss)# 变亮、变暗img_darker = darker(img)cv2.imwrite(file_dir + img_name[0:-4] + '_darker.jpg', img_darker)img_brighter = brighter(img)cv2.imwrite(file_dir + img_name[0:-4] + '_brighter.jpg', img_brighter)blur = cv2.GaussianBlur(img, (7, 7), 1.5)#      cv2.GaussianBlur(图像，卷积核，标准差）cv2.imwrite(file_dir + img_name[0:-4] + '_blur.jpg', blur)

2、标注

将jpg图片使用labelme标注或者make sence（Make Sense），我使用的后者，后者是网页版的，可以批量导入，进行标注。标注完后，导出为yolov8需要的数据格式，即txt的标注文件。

3、划分

将数据集按照训练集、测试集、验证集=8:1:1的比例进行划分，代码如下：第84、85、86行分别需要改成自己的需要划分的图片的地址、txt文件的地址，以及划分后存放的位置。运行后，我的split文件结构是split下有images和labels，这两个下级文件都含有test、train、val。

# 将图片和标注数据hatDataXml文件，按比例切分为 训练集和测试集 分割后的在split文件中
import shutil
import random
import os
import argparse# 检查文件夹是否存在
def mkdir(path):if not os.path.exists(path):os.makedirs(path)def main(image_dir, txt_dir, save_dir):# 创建文件夹mkdir(save_dir)images_dir = os.path.join(save_dir, 'images')labels_dir = os.path.join(save_dir, 'labels')img_train_path = os.path.join(images_dir, 'train')img_test_path = os.path.join(images_dir, 'test')img_val_path = os.path.join(images_dir, 'val')label_train_path = os.path.join(labels_dir, 'train')label_test_path = os.path.join(labels_dir, 'test')label_val_path = os.path.join(labels_dir, 'val')mkdir(images_dir)mkdir(labels_dir)mkdir(img_train_path)mkdir(img_test_path)mkdir(img_val_path)mkdir(label_train_path)mkdir(label_test_path)mkdir(label_val_path)# 数据集划分比例，训练集75%，验证集15%，测试集15%，按需修改train_percent = 0.8val_percent = 0.1test_percent = 0.1total_txt = os.listdir(txt_dir)num_txt = len(total_txt)list_all_txt = range(num_txt)  # 范围 range(0, num)num_train = int(num_txt * train_percent)num_val = int(num_txt * val_percent)num_test = num_txt - num_train - num_valtrain = random.sample(list_all_txt, num_train)# 在全部数据集中取出trainval_test = [i for i in list_all_txt if not i in train]# 再从val_test取出num_val个元素，val_test剩下的元素就是testval = random.sample(val_test, num_val)print("训练集数目：{}, 验证集数目：{},测试集数目：{}".format(len(train), len(val), len(val_test) - len(val)))for i in list_all_txt:name = total_txt[i][:-4]srcImage = os.path.join(image_dir, name + '.jpg')srcLabel = os.path.join(txt_dir, name + '.txt')if i in train:dst_train_Image = os.path.join(img_train_path, name + '.jpg')dst_train_Label = os.path.join(label_train_path, name + '.txt')shutil.copyfile(srcImage, dst_train_Image)shutil.copyfile(srcLabel, dst_train_Label)elif i in val:dst_val_Image = os.path.join(img_val_path, name + '.jpg')dst_val_Label = os.path.join(label_val_path, name + '.txt')shutil.copyfile(srcImage, dst_val_Image)shutil.copyfile(srcLabel, dst_val_Label)else:dst_test_Image = os.path.join(img_test_path, name + '.jpg')dst_test_Label = os.path.join(label_test_path, name + '.txt')shutil.copyfile(srcImage, dst_test_Image)shutil.copyfile(srcLabel, dst_test_Label)if __name__ == '__main__':"""python split_datasets.py --image-dir my_datasets/color_rings/imgs --txt-dir my_datasets/color_rings/txts --save-dir my_datasets/color_rings/train_data"""parser = argparse.ArgumentParser(description='split datasets to train,val,test params')parser.add_argument('--image-dir', type=str, default='/Users/jing/Documents/code/ultralytics-8.1.0/hatDataXml/JPEGImages', help='image path dir')parser.add_argument('--txt-dir', type=str, default='/Users/jing/Documents/code/ultralytics-8.1.0/hatDataXml/Annotations', help='txt path dir')parser.add_argument('--save-dir', default='/Users/jing/Documents/code/ultralytics-8.1.0//split', type=str, help='save dir')args = parser.parse_args()image_dir = args.image_dirtxt_dir = args.txt_dirsave_dir = args.save_dirmain(image_dir, txt_dir, save_dir)

划分后，在datasets下新建文件DataSet1Parts，下级目录新建test、train、val，再下级分别新建images和labels，将划分好的，按照文件名复制进去。

三、改yaml

打开ultralytics/cfg/datasets/，复制文件coco128.yaml，命名为mydata.yaml，修改路径如下，其中nc代表类别数，names代表类别名称，按照自己的数据集改即可。

# Ultralytics YOLO 🚀, AGPL-3.0 license
# COCO128 dataset https://www.kaggle.com/ultralytics/coco128 (first 128 images from COCO train2017) by Ultralytics
# Documentation: https://docs.ultralytics.com/datasets/detect/coco/
# Example usage: yolo train data=coco128.yaml
# parent
# ├── ultralytics
# └── datasets
#     └── coco128  ← downloads here (7 MB)path: ../datasets/DataSet1Parts
test: test/images
train: train/images
val: val/imagesnc: 3
#names: ['cnn','mxn','xcn']names:0: cnn1: mxn2: xcn

四、训练及验证

1、训练

在Terminal控制台输入命令：

#终端复制以下代码训练
#yolo train data=/Users/jing/Documents/code/ultralytics-8.1.0/ultralytics/cfg/datasets/mydata.yaml model=yolov8n.pt epochs=100 lr0=0.01 batch=4

2、验证

#验证测试结果，找图片放在ultralytics/assets文件夹下
#输入以下代码测试，使用效果最好的权重文件 best.pt
# yolo predict model=/Users/jing/Documents/code/ultralytics-8.1.0/runs/detect/train16/weights/best.pt source=/Users/jing/Documents/code/ultralytics-8.1.0/ultralytics/assets/ceshi.JPG