🔄Update: [Lab-5] 基于YOLOv8的工业缺陷检测实验

.gitignore

@@ -4,4 +4,5 @@
 model
 runs
 datasets
-yolov8n.pt
+yolov8n.pt
+dataset

Lab-5/src/detect.py

@@ -0,0 +1,20 @@
+from ultralytics import YOLO
+import cv2
+
+# 加载最佳模型
+model = YOLO('runs/detect/yolov8n_pcb_defect_exp14/weights/best.pt')
+
+# 在单张图像上推理
+results = model('../dataset/images/val/patches_296.jpg', save=True, conf=0.5)
+
+# 可视化结果（使用OpenCV）
+for result in results:
+    plotted_img = result.plot()  # 绘制检测结果
+    cv2.imshow('Detection Result', plotted_img)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
+
+# 支持文件夹、视频、摄像头推理
+# results = model('path/to/video.mp4', save=True, show=True)  # 视频文件
+# results = model(0, show=True)  # 摄像头（0为默认摄像头）
+

Lab-5/src/pcb_defect.yaml

@@ -0,0 +1,8 @@
+path: ../dataset  # 数据集根目录
+train: images/train      # 训练图像路径
+val: images/val          # 验证图像路径
+
+# 类别数
+nc: 6
+# 类别名称
+names: ['missing_hole', 'mouse_bite', 'open_circuit', 'short', 'spur', 'spurious_copper']

Lab-5/src/train.py

@@ -0,0 +1,16 @@
+from ultralytics import YOLO
+
+# 加载预训练模型
+model = YOLO('yolov8n.pt')  # 可选模型: yolov8n, yolov8s, yolov8m, yolov8l, yolov8x
+
+# 训练模型
+results = model.train(
+    data='pcb_defect.yaml',  # 数据集配置文件路径
+    epochs=20,              # 训练轮次
+    imgsz=640,               # 输入图像大小
+    batch=16,                # 批次大小（根据GPU内存调整）
+    name='yolov8n_pcb_defect_exp1',  # 实验名称
+    optimizer='AdamW',       # 优化器
+    lr0=0.001,               # 初始学习率
+    device='cpu'             # 使用设备（如GPU：device=0）
+)

Lab-5/src/transform_yolo.py

@@ -0,0 +1,61 @@
+import xml.etree.ElementTree as ET
+
+import pickle
+import os
+from os import listdir, getcwd
+from os.path import join
+import glob
+
+classes = ['crazing', 'inclusion', 'patches', 'pitted_surface', 'rolled-in_scale', 'scratches']  # xml文件中标记的种类
+
+
+def convert(size, box):
+    dw = 1.0 / size[0]
+    dh = 1.0 / size[1]
+    x = (box[0] + box[1]) / 2.0
+    y = (box[2] + box[3]) / 2.0
+    w = box[1] - box[0]
+    h = box[3] - box[2]
+    x = x * dw
+    w = w * dw
+    y = y * dh
+    h = h * dh
+    return (x, y, w, h)
+
+
+def convert_annotation(image_name):
+    try:
+        in_file = open('../dataset/annotations/val/' + image_name[:-3] + 'xml', encoding='utf-8')  # 原来的xml文件路径
+    except FileNotFoundError:
+        print(f"Warning: XML file for {image_name} not found, skipping.")
+        return
+
+    out_file = open('../dataset/labels/val/' + image_name[:-3] + 'txt', 'w', encoding='utf-8')  # 转换后的txt文件存放路径
+    tree = ET.parse(in_file)
+    root = tree.getroot()
+    size = root.find('size')
+    w = int(size.find('width').text)
+    h = int(size.find('height').text)
+
+    for obj in root.iter('object'):
+        difficult = obj.find('difficult').text
+        cls = obj.find('name').text
+        if cls not in classes or int(difficult) == 1:
+            continue
+        cls_id = classes.index(cls)
+        xmlbox = obj.find('bndbox')
+        b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
+             float(xmlbox.find('ymax').text))
+        bb = convert((w, h), b)
+        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
+    in_file.close()
+    out_file.close()
+
+
+wd = getcwd()
+
+if __name__ == '__main__':
+
+    for image_path in glob.glob("../dataset/images/val/*.jpg"):  # xml对应的图片的路径
+        image_name = image_path.split('\\')[-1]
+        convert_annotation(image_name)

Lab-5/src/val.py

@@ -0,0 +1,9 @@
+from ultralytics import YOLO
+
+# 加载最佳模型
+best_model = YOLO('runs/detect/yolov8n_pcb_defect_exp14/weights/best.pt')
+
+# 在测试集上评估
+metrics = best_model.val()  # 程序会自动计算mAP等指标
+print(f"mAP@0.5: {metrics.box.map50}")
+print(f"mAP@0.5:0.95: {metrics.box.map}")