阅读代码的记录

1-utils_metrics.py用在train.py中做指标衡量，现在想在推理（predict.py）的时候衡量一下指标
2-调研眼睛部位的单独分割。

https://blog.csdn.net/qq_40234695/article/details/88633094

衡量图像语义分割准确率主要有三种方法：
像素准确率（pixel accuracy, PA）
平均像素准确率（mean pixel accuracy, MPA）
平均IOU（Mean Intersection over Union, MIOU ）

像素准确率（Pixel Accuracy，PA）、
类别像素准确率（Class Pixel Accuray，CPA）、
类别平均像素准确率（Mean Pixel Accuracy，MPA）、
交并比（Intersection over Union，IoU）、
平均交并比（Mean Intersection over Union，MIoU），
其计算都是建立在混淆矩阵（Confusion Matrix）的基础上。————————————————
原文链接：https://blog.csdn.net/weixin_38353277/article/details/121029978

from ultralytics.engine.model import Model
from ultralytics.models import yolo  \# noqa
from ultralytics.nn.tasks import ClassificationModel, DetectionModel, PoseModel, SegmentationModel
class YOLO(Model):"""YOLO (You Only Look Once) object detection model."""@propertydef task_map(self):"""Map head to model, trainer, validator, and predictor classes."""return {'classify': {'model': ClassificationModel,'trainer': yolo.classify.ClassificationTrainer,'validator': yolo.classify.ClassificationValidator,'predictor': yolo.classify.ClassificationPredictor, },'detect': {'model': DetectionModel,'trainer': yolo.detect.DetectionTrainer,'validator': yolo.detect.DetectionValidator,'predictor': yolo.detect.DetectionPredictor, },'segment': {'model': SegmentationModel,'trainer': yolo.segment.SegmentationTrainer,'validator': yolo.segment.SegmentationValidator,'predictor': yolo.segment.SegmentationPredictor, },'pose': {'model': PoseModel,'trainer': yolo.pose.PoseTrainer,'validator': yolo.pose.PoseValidator,'predictor': yolo.pose.PosePredictor, }, }

YOLO类继承了ultralytics文件里面的engine文件里面的model.py文件里面的 Model类别。这个时候应该去看from ultralytics.engine.model import Model里面Model里的原码（但是下面的@property下面的代码又是什么意思呢）。

from yolov8.ultralytics import YOLO
from segment_anything.utils.transforms import ResizeLongestSide
from segment_anything import SamPredictor, sam_model_registryclass TongueSeg():def __init__(self, device = 'cuda:0', model_path="/share1/luli/yolov8SAM/pretrained_model") -> None:self.device = deviceself.model_type = 'vit_b'self.checkpoint = model_path+'/tonguesam.pth'self.det_model=YOLO('/share1/luli/yolov8/runs/detect/train20/weights/best.pt', task='detect') self.sam_model = sam_model_registry[self.model_type](checkpoint=self.checkpoint).to(device)

   self.det_model=YOLO('/share1/luli/yolov8/runs/detect/train20/weights/best.pt', task='detect') 是from yolov8.ultralytics import YOLO，导入YOLO--->>>点击继续入YOLO，进入到下面这个地方：from ultralytics.engine.model import Modelfrom ultralytics.models import yolo  # noqafrom ultralytics.nn.tasks import ClassificationModel, DetectionModel, PoseModel, SegmentationModelclass YOLO(Model):上面又是继承了Model，Model是从ultralytics.engine.model里面导入的，这时候要点击Model去查看源码，如下：import inspectimport sysfrom pathlib import Pathfrom typing import Unionfrom ultralytics.cfg import TASK2DATA, get_cfg, get_save_dirfrom ultralytics.hub.utils import HUB_WEB_ROOTfrom ultralytics.nn.tasks import attempt_load_one_weight, guess_model_task, nn, yaml_model_loadfrom ultralytics.utils import ASSETS, DEFAULT_CFG_DICT, LOGGER, RANK, callbacks, checks, emojis, yaml_load		class Model(nn.Module):上面导入了cfg,nn.tasks,utils

utils里面有个__init__.py函数，函数上面有个callbacks文件夹，文件夹里面有很多的结构可视化