Add files via upload

Author: Shengxiang-Lin

Key Function/DataSet.py

@@ -0,0 +1,63 @@
+# 导入相关库
+import os # 与系统文件交互
+import tifffile as tiff #读取tiff文件格式
+from PIL import Image #图片处理
+#与torch 相关的库
+import torch
+from torchvision import transforms
+from torch.utils.data import Dataset, DataLoader
+
+#from sklearn.preprocessing import MinMaxScaler
+import numpy as np
+
+#创建数据集
+class myDataSet(Dataset):
+    def __init__(self,img_dir,group_size=10000,transform=None):
+        self.img_dir=img_dir
+        self.images=os.listdir(img_dir)
+        self.transform=transform
+        self.all_imgs=[]
+        self.emcal=[]
+        self.hcal=[]
+        self.trkn=[]
+        self.trkp=[]
+        self.truth=[]
+        self.group_size=group_size
+        self.load_images()
+    
+    def load_images(self):
+        all_imgs=[]
+        for filename in self.images:
+            if filename.endswith(".tiff"):
+                img_path=os.path.join(self.img_dir, filename)
+            img_array=tiff.imread(img_path)
+            img=Image.fromarray(img_array)
+            img_tensor=transform(img)
+            all_imgs.append(img_tensor)
+        self.emcal=all_imgs[:self.group_size]
+        self.hcal=all_imgs[self.group_size:2*self.group_size]
+        self.trkn=all_imgs[2*self.group_size:3*self.group_size]
+        self.trkp=all_imgs[3*self.group_size:4*self.group_size]
+        self.truth=all_imgs[4*self.group_size:5*self.group_size]
+        
+        self.X=[]
+        self.Y=[]
+        
+        for emcal, hcal, trkn, trkp in zip(self.emcal,self.hcal,self.trkn, self.trkp):
+            combined_features=torch.stack((emcal,hcal,trkn,trkp))
+            self.X.append(combined_features)
+        self.X=torch.stack(self.X).squeeze()
+        self.Y=torch.stack(self.truth).squeeze()
+        
+    def __len__(self):
+        return len(self.X)
+    def __getitem__(self,idx):
+        return self.X[idx], self.Y[idx]
+transform=transforms.Compose([
+    transforms.ToTensor(),
+    # 数据预处理后期添加
+])
+    
+
+
+

Key Function/Plotter.py

@@ -0,0 +1,92 @@
+#绘制直方图模块
+import numpy as np
+import matplotlib.pyplot as plt
+import matplotlib.colors as mcolors
+class Plotter:
+    def __init__(self):
+        pass
+    def plot_histogram(self, data_list, num_bins=10, color='blue', title='Histogram',xlabel='Values',ylabel='Frequency',net = 'None',
+                    text = 'None',need_log_scale=False,range=[-1,-1],image_info = 'None',data_points='0'):
+        # 计算范围
+        data_min, data_max = np.min(data_list), np.max(data_list)
+        if range == [-1,-1]:
+            data_range = (data_min, data_max)
+        else :
+            data_range = range
+        hist, bin_edges=np.histogram(data_list,bins=num_bins,range=data_range)
+        max_bin_index = np.argmax(hist)
+        max_bin_range = (bin_edges[max_bin_index], bin_edges[max_bin_index + 1])
+        max_bin_freq = hist[max_bin_index] # 该 bin 的频率
+        print(f"频率最高的 bin 的范围是: {max_bin_range}，频率为: {max_bin_freq}")
+        plt.hist(data_list,bins=num_bins,range=data_range,histtype='stepfilled',align='mid',orientation='vertical',color=color)
+        if need_log_scale:
+            plt.yscale('log')
+        plt.title(title,fontsize=17)
+        plt.xlabel(xlabel,fontsize=12)
+        plt.tick_params(labelsize=12)
+        plt.ylabel(ylabel,fontsize=12)
+        #在图中添加文本
+        plt.text(0.5, -0.2, f"The most frequent data bin: {max_bin_range}\nThe std of the data is {np.std(data_list)}",
+                ha='center', va='center', transform=plt.gca().transAxes,fontsize=12)
+        if text != 'None':
+            plt.text(0.5, -0.3, text, ha='center', va='center', transform=plt.gca().transAxes,fontsize=12)
+        if image_info != 'None':
+            plt.text(0.5, -0.4, 'Generated from '+image_info+ '. The num of data points is '+ data_points, ha='center', va='center', 
+                    transform=plt.gca().transAxes,fontsize=12)
+        if net != 'None':
+            plt.text(0.5, -0.5, 'The network is '+net, ha='center', va='center', 
+                    transform=plt.gca().transAxes,fontsize=12)
+    
+    def plot_scatter(self,list1,list2,title='scatter',color='r',s=1,xlabel='xlabel',ylabel='ylabel', text = 'None',net = 'None',
+                    image_info = 'None',data_points='0'):
+        fig, ax1=plt.subplots(figsize=(10,10))
+        # 绘制对角线
+        ax1.plot([min(list1), max(list1)], [min(list1), max(list1)], 'k--', label='X=Y')
+        # 绘制散点图
+        ax1.scatter(list1, list2, s=s, color=color)
+        ax1.set_xlabel(xlabel)
+        ax1.set_ylabel(ylabel)
+        plt.tick_params(labelsize=12)
+        ax1.set_title(title)
+        # 添加文本
+        plt.text(0.5, -0.1, text, ha='center', va='center', transform=plt.gca().transAxes)
+        # 拟合散点图
+        coefficients = np.polyfit(list1, list2, 1)
+        polynomial = np.poly1d(coefficients)
+        x_fit = np.linspace(min(list1), max(list1), 100)
+        y_fit = polynomial(x_fit)
+        if text != 'None':
+            plt.text(0.5, -0.3, text, ha='center', va='center', transform=plt.gca().transAxes,fontsize=12)
+        if image_info != 'None':
+            plt.text(0.5, -0.2, 'Generated from '+image_info+ '. The num of data points is '+ data_points, ha='center', va='center', 
+                    transform=plt.gca().transAxes,fontsize=12)
+        if net != 'None':
+            plt.text(0.5, -0.4, 'The network is '+net, ha='center', va='center', transform=plt.gca().transAxes,fontsize=12)
+        # 绘制最小二乘拟合线
+        ax1.plot(x_fit, y_fit, 'r-', label='Least Squares Fit', linewidth=2, alpha=0.7,color='b')
+        
+    def plot_error_pixel_map(self, true_list, predict_list, HIGTHT=56, WIDTH=56,text = 'None',net='None',
+                    image_info = 'None',data_points='0'):
+        # 计算误差矩阵
+        Error_pixel_map = np.array(true_list) - np.array(predict_list)
+        Error_pixel_map = Error_pixel_map.astype(np.float32).reshape(HIGTHT, WIDTH)
+        # 创建颜色映射
+        plt.tick_params(labelsize=12)
+        cmap = mcolors.LinearSegmentedColormap.from_list('custom_cmap', ['red', 'white', 'black'])
+        norm = mcolors.TwoSlopeNorm(vmin=Error_pixel_map.min(), vcenter=0, vmax=Error_pixel_map.max())
+
+        # 绘制误差像素图
+        plt.figure(figsize=(10, 5))
+        plt.imshow(Error_pixel_map, cmap=cmap, norm=norm)
+        plt.colorbar(label='Error Value')
+        plt.title('Error Pixel Map')
+        if text != 'None':
+            plt.text(0.5, -0.1, text, ha='center', va='center', transform=plt.gca().transAxes,fontsize=12)
+        if image_info != 'None':
+            plt.text(0.5, -0.2, 'Generated from '+image_info+ '. The num of data points is '+ data_points, ha='center', va='center', 
+                    transform=plt.gca().transAxes,fontsize=12)
+        if net != 'None':
+            plt.text(0.5, -0.3, 'The network is '+net, ha='center', va='center', 
+                    transform=plt.gca().transAxes,fontsize=12)
+        plt.tight_layout()
+        plt.show()