# env_deploy(無GUP環境) ### 下載YOLO v7 ``` cmd git clone https://github.com/WongKinYiu/yolov7 ``` ### 下載權重檔，放入yolov7資料夾內 https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt ### 打開CMD，先把調整到YOLO v7的目錄後，輸入以下指令，安裝相關套件 ``` cd yolov7 pip install -r requirements.txt ``` ### 執行以下指令來測試 將原始圖片放在 inference/images 中 執行以下指令即會在runs\detect\exp看到結果 ``` cmd python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source inference/images/horses.jpg --view-img ``` # 訓練自訂義模型 ### 1.以下目錄創建yolov7_custom.yaml C:\Users\michelin.jhong\Desktop\Yolo\keras_yolo\yolov7\cfg\training\yolov7_custom.yaml nc為模型的類別數量 ![](https://hackmd.io/_uploads/Byf8Ng0qh.png) ### 2.模型分類 以及 類別標籤 的設定 C:\Users\michelin.jhong\Desktop\Yolo\keras_yolo\yolov7\data\mydata.yaml 模型分類 以及 類別標籤 的設定 ![](https://hackmd.io/_uploads/rytDNl093.png) ### 3.資料準備 以下為資料存放路徑 ``` C:\Users\michelin.jhong\Desktop\Yolo\keras_yolo\yolov7\mydataset 需使用標註程式ex.makescene ``` (一圖片 + .txt檔) 為YOLO格式 ![](https://hackmd.io/_uploads/ByAooLPqh.png) ### 4.為將資料集分割成訓練驗證 ``` C:\Users\michelin.jhong\Desktop\Yolo\keras_yolo\yolov7\mydataset\程式.py n n 為要取幾筆資料為驗證資料 ``` ```python #程式.py import os import sys import shutil from random import sample from collections import OrderedDict if len(sys.argv) < 2: print('input val num') exit() valNum = int(sys.argv[1]) lst = os.listdir('all') lst.remove('classes.txt') for f in lst: if 'txt' not in f: extension = f.split('.')[-1] break names = [i.split('.')[0] for i in lst] names = list(OrderedDict.fromkeys(names)) valNames = sorted(sample(names, valNum)) names = sorted(list(set(names).difference(set(valNames)))) paths = [os.path.join('images', 'train'), os.path.join('images', 'val'), os.path.join('labels', 'train'), os.path.join('labels', 'val')] for p in paths: os.makedirs(p) trainPath = [] for fname in names: orgImgPath = os.path.join('all', f'{fname}.{extension}') newImgPath = os.path.join(paths[0], f'{fname}.{extension}') trainPath.append(os.path.abspath(newImgPath) + '\n') orgTxtPath = os.path.join('all', f'{fname}.txt') newTxtPath = os.path.join(paths[2], f'{fname}.txt') shutil.copy(orgImgPath, newImgPath) shutil.copy(orgTxtPath, newTxtPath) valPath = [] for fname in valNames: orgImgPath = os.path.join('all', f'{fname}.{extension}') newImgPath = os.path.join(paths[1], f'{fname}.{extension}') valPath.append(os.path.abspath(newImgPath) + '\n') orgTxtPath = os.path.join('all', f'{fname}.txt') newTxtPath = os.path.join(paths[3], f'{fname}.txt') shutil.copy(orgImgPath, newImgPath) shutil.copy(orgTxtPath, newTxtPath) with open('train.txt', 'w') as f: f.writelines(trainPath) with open('val.txt', 'w') as f: f.writelines(valPath) shutil.copy(os.path.join('all', 'classes.txt'), 'classes.names') ``` ``` 執行上述程式最終會形成右下角的樣子 ``` ![](https://hackmd.io/_uploads/r1Uo7lR9h.jpg) # 使用colab 訓練 ## 掛載colab ```py from google.colab import drive drive.mount('/content/drive') ``` ## 切換位置 ```python %cd /content/drive/MyDrive/yolov7/yolov7 ``` ## 環境設定 Env Create ```python !pip install --upgrade setuptools pip --user !pip install onnx !pip install coremltools>=4.1 ``` ### 查看版本 ```py import sys import torch print(f"Python version: {sys.version}, {sys.version_info} ") print(f"Pytorch version: {torch.__version__} ") ``` ### clone 專案 ```python !nvidia-smi !# Download YOLOv7 code !git clone https://github.com/WongKinYiu/yolov7 %cd yolov7 !ls ``` ### 下載訓練權重 ```py Download trained weights: !wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt ``` ### 物件偵測 ```py !python detect.py --weights ./yolov7.pt --conf 0.25 --img-size 640 --source inference/images/horses.jpg ``` ### 訓練模型 ```python= !python train.py --workers 8 --device 0 --batch-size 8 --data ./data/mydata.yaml --img 640 640 --cfg ./cfg/training/yolov7_custom.yaml --hyp ./data/hyp.scratch.p5.yaml --weights ./yolov7.pt ``` ``` --cfg 接受模型配置的参数 --data 接收數據配置的参数 --device 0 訓練類型，我是一塊GPU 所以用0 --batch-size 8 GPU内存大小决定 --epoch 訓練次數，建議300 --weights 訓練權重 ``` ### 使用模型 ```python= 指令： python detect.py --weights yolov7x.pt --conf 0.25 --img-size 640 --source inference/images/ –weights :预权重路径 (detect.py的相对路径) –source: 需训练的文件路径 (detect.py的相对路径) –img-size:输入网络图像大小，默认640 * 640 –project:训练模型保存的位置，默认为run/train –name: 保存项目名字，一般是run/train/exp –save-txt :选择是否生成结果文本 –save_conf:结果文本保存相似度 ``` # 其他功能 ## 輸出檢測到的目標坐標信息/位置 ```python 找到detect.py，在大概113行，找到plot_one_box ctr+鼠标點擊，進入general.py，並自動定位到plot_one_box函数，修改函数为 #--------------------------------------------------------------------- def plot_one_box(x, img, 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) print("左上角的坐标为：(" + str(c1[0]) + "," + str(c1[1]) + ")，右下角的坐标为(" + str(c2[0]) + "," + str(c2[1]) + ")") ``` ### 注意 ``` colab訓練的路徑要更改，還有 train.txt & val.txt 中的圖片路徑要設定正確 ``` # Reference： https://ithelp.ithome.com.tw/articles/10306298 https://blog.eddie.tw/category/yolo-v7/ https://github.com/WongKinYiu/yolov7