基于YOLO26的微塑料缺陷检测系统（Python源码+数据集+Pyside6界面）

重复模块m (n次迭代):
┌─────────────────────────────────────────────────────────┐
│                                                         │
│  如果 attn=True:                                        │
│    Sequential(                                          │
│        Bottleneck(self.c, self.c),                     │  ←─ 先特征提取
│        PSABlock(self.c, attn_ratio=0.5, num_heads=...) │  ←─ 后注意力增强
│    )                                                    │
│                                                         │
│  否则如果 c3k=True:                                     │
│    C3k(self.c, self.c, 2)                             │  ←─ 同YOLOv11
│                                                         │
│  否则:                                                  │
│    Bottleneck(self.c, self.c)                         │  ←─ 同YOLOv11
│                                                         │
└─────────────────────────────────────────────────────────┘

class BboxLoss(nn.Module):
    """Criterion class for computing training losses for bounding boxes."""

    def __init__(self, reg_max: int = 16):
        """Initialize the BboxLoss module with regularization maximum and DFL settings."""
        super().__init__()
        self.dfl_loss = DFLoss(reg_max) if reg_max > 1 else None

# microplastic_yolo.py
# 微塑料二分类 YOLO 检测伪代码：microfiber / microfragment

from pathlib import Path
import yaml
from ultralytics import YOLO

# ==================== 配置 ====================
DATASET_DIR = Path("microplastic-2class-yolo")
DATA_YAML = DATASET_DIR / "data.yaml"

MODEL_WEIGHT = "yolov8n.pt"
PROJECT = "runs/microplastic"
RUN_NAME = "train"
BEST_WEIGHT = Path(PROJECT) / RUN_NAME / "weights" / "best.pt"

CLASS_NAMES = {
    0: "microfiber",
    1: "microfragment",
}


# ==================== 生成 data.yaml ====================
def create_data_yaml():
    data = {
        "path": str(DATASET_DIR.resolve()),
        "train": "images/train",
        "val": "images/val",
        "names": CLASS_NAMES,
    }
    with open(DATA_YAML, "w", encoding="utf-8") as f:
        yaml.safe_dump(data, f, allow_unicode=True)
    print(f"已生成: {DATA_YAML}")


# ==================== 训练 ====================
def train():
    create_data_yaml()
    model = YOLO(MODEL_WEIGHT)
    model.train(
        data=str(DATA_YAML),
        epochs=100,
        imgsz=640,
        batch=16,
        project=PROJECT,
        name=RUN_NAME,
    )


# ==================== 验证 ====================
def validate():
    model = YOLO(str(BEST_WEIGHT))
    metrics = model.val(data=str(DATA_YAML))
    print("mAP50:", metrics.box.map50)
    print("mAP50-95:", metrics.box.map)


# ==================== 推理并计数 ====================
def detect_and_count(source="test/images"):
    model = YOLO(str(BEST_WEIGHT))
    results = model.predict(
        source=source,
        conf=0.25,
        iou=0.5,
        save=True,
    )

    counts = {name: 0 for name in CLASS_NAMES.values()}

    for r in results:
        for cls_id in r.boxes.cls:
            name = model.names[int(cls_id)]
            counts[name] += 1

    print("检测统计：", counts)
    return results


# ==================== 主程序 ====================
if __name__ == "__main__":
    # 首次运行：训练模型
    # train()

    # 验证模型
    # validate()

    # 推理并统计微纤维、微碎片数量
    detect_and_count("test/images")

names:
  0: 微塑料纤维
  1: 微塑料碎片

path: D:/YOLO26/data/microplastic

train: train/images
val: valid/images
test: test/images

# Number of classes
nc: 2

# Class names (index -> name).
names:
  0: microfiber
  1: microfragment

import warnings
warnings.filterwarnings('ignore')
from ultralytics import YOLO

if __name__ == '__main__':
    model = YOLO('ultralytics/cfg/models/26/yolo26.yaml')
    model.train(data='data/microplastic.yaml',
                cache=False,
                imgsz=640,
                epochs=200,
                batch=8,
                close_mosaic=10,
                device='0',
                optimizer='SGD', # using SGD
                project='runs/train',
                name='exp',
                )

YOLO26 summary (fused): 122 layers, 2,375,226 parameters, 0 gradients, 5.3 GFLOPs
                 Class     Images  Instances      Box(P          R      mAP50  mAP50-95): 100% ━━━━━━━━━━━━ 30/30 3.1it/s 9.6s
                   all        957       6642      0.881       0.89       0.93      0.595
            microfiber        714       2332      0.881      0.898      0.943      0.699
         microfragment        629       4310      0.882      0.882      0.917      0.491

pip install --upgrade pip
pip install pyside6 -i https://mirror.baidu.com/pypi/simple