准备环境

1. 在github上打开https://github.com/ultralytics/ultralytics
2. 点击 跳转到Colab
3. 点击复制到云端硬盘，将运行环境复制到自己的云盘
4. 安装yolov8

   %pip install ultralytics
   import ultralytics
   ultralytics.checks()

准备数据

1. 首先数据集压缩包上传到谷歌云盘

2. 挂载云端硬盘

   from google.colab import drive
   drive.mount('/content/drive')

3. 解压数据集

   !unzip /content/drive/MyDrive/defect_Images.zip -d /content/dataset

4. 将数据集转化为yolo格式，并生成训练集和验证集。此处使用的数据集是天池比赛：2019广东工业智造创新大赛 赛场一的数据集，注意修改josn_path和image_path。

   import numpy as np
   import os
   import json
   from tqdm.auto import tqdm
   import shutil as sh
   import cv2
   josn_path = "/content/drive/MyDrive/Annotations/anno_train.json"
   image_path = "/content/dataset/defect_Images"
   defect_name2label = {
   '破洞': 1, '水渍': 2, '油渍': 2, '污渍': 2, '三丝': 3, '结头': 4, '花板跳': 5, '百脚': 6, '毛粒': 7,
   '粗经': 8, '松经': 9, '断经': 10, '吊经': 11, '粗维': 12, '纬缩': 13, '浆斑': 14, '整经结': 15, '星跳': 16, '跳花': 16,
   '断氨纶': 17, '稀密档': 18, '浪纹档': 18, '色差档': 18, '磨痕': 19, '轧痕': 19, '修痕': 19, '烧毛痕': 19, '死皱': 0, '云织': 0,
   '双纬': 0, '双经': 0, '跳纱': 0, '筘路': 0, '纬纱不良': 0,
   }
   name_list = []
   image_h_list = []
   image_w_list = []
   c_list = []
   w_list = []
   h_list = []
   x_center_list = []
   y_center_list = []
   with open(josn_path, 'r') as f:
   temps = tqdm(json.loads(f.read()))
   for temp in temps:
       # image_w = temp["image_width"]
       # image_h = temp["image_height"]
       name = temp["name"].split('.')[0]
       path = os.path.join(image_path, temp["name"])
       # print('path: ',path)
       im = cv2.imread(path)
       sp = im.shape
       image_h, image_w = sp[0], sp[1]
       # print("image_h, image_w: ", image_h, image_w)
       # print("defect_name: ",temp["defect_name"])
       # bboxs
       x_l, y_l, x_r, y_r = temp["bbox"]
       # print(temp["name"], temp["bbox"])
       defect_name = defect_name2label[temp["defect_name"]]
       # print(image_w, image_h)
       # print(defect_name)
       x_center = (x_l + x_r) / (2 * image_w)
       y_center = (y_l + y_r) / (2 * image_h)
       w = (x_r - x_l) / (image_w)
       h = (y_r - y_l) / (image_h)
       # print(x_center, y_center, w, h)
       name_list.append(temp["name"])
       c_list.append(defect_name)
       image_h_list.append(image_w)
       image_w_list.append(image_h)
       x_center_list.append(x_center)
       y_center_list.append(y_center)
       w_list.append(w)
       h_list.append(h)
   index = list(set(name_list))
   print(len(index))
   for fold in [0]:
       val_index = index[len(index) * fold // 5:len(index) * (fold + 1) // 5]
       print(len(val_index))
       for num, name in enumerate(name_list):
           print(c_list[num], x_center_list[num], y_center_list[num], w_list[num], h_list[num])
           row = [c_list[num], x_center_list[num], y_center_list[num], w_list[num], h_list[num]]
           if name in val_index:
               path2save = 'val/'
           else:
               path2save = 'train/'
           # print('convertor\\fold{}\\labels\\'.format(fold) + path2save)
           # print('convertor\\fold{}/labels\\'.format(fold) + path2save + name.split('.')[0] + ".txt")
           # print("{}/{}".format(image_path, name))
           # print('convertor\\fold{}\\images\\{}\\{}'.format(fold, path2save, name))
           if not os.path.exists('convertor/fold{}/labels/'.format(fold) + path2save):
               os.makedirs('convertor/fold{}/labels/'.format(fold) + path2save)
           with open('convertor/fold{}/labels/'.format(fold) + path2save + name.split('.')[0] + ".txt", 'a+') as f:
               for data in row:
                   f.write('{} '.format(data))
               f.write('\n')
               if not os.path.exists('convertor/fold{}/images/{}'.format(fold, path2save)):
                   os.makedirs('convertor/fold{}/images/{}'.format(fold, path2save))
               sh.copy(os.path.join(image_path, name),
                       'convertor/fold{}/images/{}/{}'.format(fold, path2save, name))

5. 修改配置文件my.yaml

   # Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
   path: /content/convertor/fold0  # dataset root dir
   train: images/train  # train images (relative to 'path') 128 images
   val: images/val  # val images (relative to 'path') 128 images
   test:  # test images (optional)
   # Classes
   names:
   0: 0
   1: 1
   2: 2
   3: 3
   4: 4
   5: 5
   6: 6
   7: 7
   8: 8
   9: 9
   10: 10
   11: 11
   12: 12
   13: 13
   14: 14
   15: 15
   16: 16
   17: 17
   18: 18
   19: 19

附：数据集转换工具https://github.com/FireworksFallDown/TypeTotype，可将COCO，YOLO，VOC数据集相互转化

模型训练

可用以下命令行方式进行训练

!yolo detect train model=yolov8n.yaml data=/content/drive/MyDrive/my.yaml epochs=200 batch=64 imgsz=640

训练参数详解：

模型验证

使用如下命令，即可完成对验证数据的评估。

!yolo detect val model=runs/detect/train3/weights/best.pt data=/content/drive/MyDrive/my.yaml

验证参数详解：

模型预测

使用如下命令，即可完成对新数据的预测，source需要指定为自己的图像路径，或者摄像头（0）。

!yolo detect predict model=runs/detect/train3/weights/best.pt source=data/images

预测参数详解：

模型导出

使用如下命令，即可完成训练模型的导出。

!yolo detect export model=runs/detect/train3/weights/best.pt