# tile_defect_and_color_detect

**Repository Path**: hehuang0724/tile_defect_and_color_detect

## Basic Information

- **Project Name**: tile_defect_and_color_detect
- **Description**: 瓷砖缺陷和色差检测软件
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-03-27
- **Last Updated**: 2025-02-06

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README


# 使用
## 添加模型

### 1. JSON配置
在qt/DetectModel/model.json 中添加
```json
{
  "defect_model": [
    {
      "name": "yolov8",
      "weight_path": "",
      "func_path": "",
      "type": ""
    }
  ],
  "abberration_model": [
    {
      "name": "resnet50",
      "weight_path": "",
      "func_path": "",
      "type": ""
    }
  ]
}
```

注释：主要实现name字段，其他字段暂不需要

### 2. 添加模型（源码方式）
- 工作目录为
    + qt/DetectModel/DefectModel
    + qt/DetectModel/AberrationModel

- 以添加YOLOv8为例
1. 在qt/DetectModel/DefectModel下新建yolov8/source文件夹
2. 新建detect_logical.py文件，新建类Yolov8继承BaseModel，并实现__init__、model_init、button_image_open三个方法
```python
from qt.common.core.BaseModel import BaseModel

class Yolov8(BaseModel):
    def __init__(self, ext_info):
        '''
        :param ext_info 额外信息（model.json里面的值）
        '''
        self.cap = cv2.VideoCapture()
        self.num_stop = 1 # 暂停与播放辅助信号，note：通过奇偶来控制暂停与播放
        self.output_folder = 'output/'
        self.vid_writer = None

        # 权重初始文件名
        self.openfile_name_model = ext_info['defect_model']['weight_path']

        self.ext_info = ext_info
        self.image = None

    # 加载相关参数，并初始化模型
    def model_init(self):
        pass

    # 打开图片并检测
    def button_image_open(self, image, win, win_info, detect_info):
        '''
        :param image 拍摄图像矩阵（cv.BGR）
        :param win qt展示图像界面句柄
        :param win_info qt展示检测信息界面句柄
        :param detect_info 额外展示信息前缀
        '''
        name_list = []
        # image = cv2.imread(r"D:\hakvision\hkpy_4.2.1\hk-pyqt-tile\qt\DetectModel\DefectModel\yolov5\source\data\images\25_4914_1638_1024_1024_0_8192_6000.jpg")
        if image is not None:
            # img = cv2.imread(img_name)
            info_show, img = self.dvide_slice(image)
            # info_show = "测试"
            # img = image
            # info_show = self.detect(name_list, img)
            print(info_show)
            # 获取当前系统时间，作为img文件名
            now = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime(time.time()))
            new_filename = now + '.jpg'
            file_path = self.output_folder + 'img_output/' + new_filename
            # cv2.imwrite(file_path, img)

            # 检测信息显示在界面
            detect_info.message = detect_info.message + "\n缺陷检测信息:\n" + info_show
            win_info.append("{}".format(info_show))

            # 检测结果显示在界面
            result = cv2.cvtColor(img, cv2.COLOR_BGR2BGRA)
            w, h = result.shape[1], result.shape[0]
            t_w, t_h =win.width(), win.height()
            result = cv2.resize(result, (int(t_h / h * w), t_h), interpolation=cv2.INTER_AREA)
            pad_w = t_w - result.shape[1]  # 计算需要填充的像素数目（图像的宽这一维度上）
            pad_h = t_h - result.shape[0]  # 计算需要填充的像素数目（图像的高这一维度上）
            top, bottom = pad_h // 2, pad_h - (pad_h // 2)
            left, right = pad_w // 2, pad_w - (pad_w // 2)
            result = cv2.copyMakeBorder(result, top, bottom, left, right, cv2.BORDER_CONSTANT, None, (0, 0, 0))
            QtImg = QtGui.QImage(result.data, result.shape[1], result.shape[0],
                                 QtGui.QImage.Format_RGB32)
            win.setPixmap(QtGui.QPixmap.fromImage(QtImg))
            win.setScaledContents(True)  # 设置图像自适应界面大小

            # 保存图片
            save_root = r"D:\hakvision\hkpy_4.2.1\hk-pyqt-tile\qt\Source\Save"
            detect_name = str(time.strftime("%Y-%m-%d_%H-%M-%S", time.localtime()))
            try:
                cv2.imwrite(os.path.join(save_root, "detect_{}.jpg".format(detect_name)), result)
            except:
                print(os.path.join(save_root, "detect_{}.jpg".format(detect_name)))
```
3. 在qt/DetectModel/DefectModel/yolov8新建__init__.py
```python
from .source.detect_logical import Yolov8

__all__ = ('Yolov8')
```


### 3.在ModelFactory类中注册
1. 导入具体检测类
2. 在qt/DetectModel/ModelFactory.py 中新建判断语句
```python
from .DefectModel.yolov8 import Yolov8

class ModelFactory():
    def getModel(self, ext_info):
        message = ''
        if False: pass
        elif str(ext_info['defect_model']['name']).startswith('yolov8'):
            model = Yolov8(ext_info)
            model.model_init()
            return model
```

### 4. fileName.json

该文件内保存了权重路径和检测后要保存的图片路径