73 lines
4.3 KiB
Python
73 lines
4.3 KiB
Python
# 在import部分添加median_filter
|
|
from Image_Process.image_enhancement import histogram_equalization, adaptive_histogram_equalization_without_limit, unsharp_mask, laplacian_sharpen, adjust_hsv, gamma_correction, Contrast_Limited_Adaptive_Histogram_Equalization, Hight_Light, mean_filter, median_filter
|
|
import datetime
|
|
|
|
Image_Enhance = {
|
|
"Shapen" : laplacian_sharpen,
|
|
"CLAHE": histogram_equalization,
|
|
"CLAHE_Adaptive" : adaptive_histogram_equalization_without_limit,
|
|
"CLAHE_Adaptive_have_Limit" : Contrast_Limited_Adaptive_Histogram_Equalization,
|
|
"HSV" : adjust_hsv,
|
|
"gamma" : gamma_correction,
|
|
"Hight_Light" : Hight_Light,
|
|
"Mean" : mean_filter,
|
|
"Median" : median_filter, # 添加中值濾波
|
|
"Gamma_Value" : 1.0
|
|
}
|
|
|
|
Loading_Config = {
|
|
"Test_Data_Root": "../Dataset/Testing",
|
|
"Train_Data_Root": "../Dataset/Training",
|
|
"Annotation_Training_Root": "../Dataset/Annotation/Training",
|
|
"Annotation_Testing_Root": "../Dataset/Annotation/Testing",
|
|
"TestProcess_Image_Root" : "../TestProcess_Image",
|
|
"ImageGenerator_Data_Root": "../Dataset/ImageGenerator",
|
|
"Process_Roots" : "../Dataset/test_Images",
|
|
# "Image enhance processing save root": f'../Dataset/image_enhancement_Result/New_{Image_Enhance["gamma"].__name__}_and_gamma_value_is_{Image_Enhance["Gamma_Value"]}',
|
|
"Image enhance processing save root": f'../Dataset/image_enhancement_Result/Resetting the training and testing dataset',
|
|
"Training_Labels": ["stomach_cancer_Crop", "Normal_Crop", "Have_Question_Crop"],
|
|
"Label_Image_Labels" : ["CA", "Have_Question"],
|
|
"XML_Loading_Label" : ["stomach_cancer_Crop", "Have_Question_Crop"],
|
|
"Identification_Label_Length" : 2,
|
|
}
|
|
|
|
Training_Config = {
|
|
"Model_Name": "Xception and GastoSegNet",
|
|
"CA_Experiment_Name": f"New architecture of Xception to CA and Have Question",
|
|
"Normal_Experiment_Name": f"New architecture of Xception to Normal and Others",
|
|
"Mask_Experiment_Name" : "New architecture of GastoSegNet",
|
|
"Epoch": 10000,
|
|
"Train_Batch_Size": 64,
|
|
"Image_Size": 256,
|
|
"Class_Count": 904,
|
|
"Get_Generator": "True",
|
|
"weight_decay": 0.01,
|
|
"Number_Of_Classes" : len(Loading_Config["Training_Labels"])
|
|
}
|
|
|
|
Model_Config = {
|
|
"Model Name": "xception",
|
|
"GPA Output Nodes": 2048,
|
|
"Linear Hidden Nodes": 1025,
|
|
"Output Linear Nodes": 2,
|
|
"Dropout Rate": 0.6
|
|
}
|
|
|
|
Save_Result_File_Config = {
|
|
"Identification_Plot_Image" : f"../Result/Training_Image/save_the_train_image({str(datetime.date.today())})", # 分類模型的走勢圖存檔路徑
|
|
"Segument_Plot_Image" : f"../Result/Training_Image/save_the_train_image({str(datetime.date.today())})/Segument_Plot_Image", # 分割模型的走勢圖存檔路徑
|
|
|
|
"Identification_Marix_Image" : f"../Result/Matrix_Image/model_matrix_image({str(datetime.date.today())})/Identification_Plot_Marix_Image", # 分類模型的混淆矩陣存檔路徑
|
|
|
|
"Identification_Every_Fold_Training_Result" : f'../Result/Training_Result/save_the_train_result({str(datetime.date.today())})/Identification_Every_Fold', # 分類模型每折訓練結果存檔路徑
|
|
"Identification_Average_Result" : f'../Result/Training_Average_Result/Average_Result({str(datetime.date.today())})/Identification_Average_Result', # 分類模型平均訓練結果存檔路徑
|
|
|
|
"Segument_Every_Fold_Training_Result" : f'../Result/Training_Result/save_the_train_result({str(datetime.date.today())})/Segument_Every_Fold', # 分割模型每折訓練結果存檔路徑
|
|
"Segument_Average_Result" : f'../Result/Training_Average_Result/Average_Result({str(datetime.date.today())})/Segument_Average_Result', # 分割模型平均訓練結果存檔路徑
|
|
"Segument_Bounding_Box_Image" : f'../Result/Bounding_Box_Image/save_bounding_box_image({str(datetime.date.today())})', # 分割模型邊界框圖像存檔路徑
|
|
"Segument_Test_Bounding_Box_Image" : f'../Result/Test_Bounding_Box_Image/save_bounding_box_image({str(datetime.date.today())})', # 分割模型邊界框圖像存檔路徑
|
|
|
|
"Normal_Identification_Best_Model" : '../Result/save_the_best_model/Identification_Normal',
|
|
"CA_Identification_Best_Model" : "../Result/save_the_best_model/Identification_CA",
|
|
"Segmentation_Best_Model" : "../Result/save_the_best_model/Segmentation",
|
|
} |