python读取matlab数据(.m文件 .mat文件)

转x33g5p2x  于2022-04-17 转载在 Python  
字(9.4k)|赞(0)|评价(0)|浏览(571)

首先导入scipy的包 from scipy.io import loadmat
然后读取 m = loadmat(“F:/__identity/activity/论文/data/D001.mat”)
注意这里m是一个dict数据结构

4.接下来就是用Python读取上一步中保存的matlab工作区的数据Data。Python中我们需要用到scipy库，这里我们先import进去

import scipy.io as scio

data=scio.loadmat('./matlab.mat')

type(data)

输出的为dict字典类型

7.读取对应我们想要的数据

这里我们假设需要将数据matlab_y读进python中（这里我们用numpy库将数据转化为数组类型）

import numpy as np #导入矩阵处理库
python_y=np.array(data['matlab_y']) #将matlab数据赋值给python变量

至此，就完成了使用python读取matlab数据。
————————————————
版权声明：本文为CSDN博主「刚开始的人生」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/qq_44946715/article/details/119932640

表情识别的处理：

https://github.com/chenxindaaa/emotic/blob/main/mat2py.py

  1. import argparse 
  2. import csv 
  3. import cv2
  4. import numpy as np 
  5. import os 
  6. from scipy.io import loadmat 
  8. class emotic_train:
  9.     def __init__(self, filename, folder, image_size, person):
  10.         self.filename = filename
  11.         self.folder = folder
  12.         self.im_size = []
  13.         self.bbox = []
  14.         self.cat = []
  15.         self.cont = []
  16.         self.gender = person[3][0]
  17.         self.age = person[4][0]
  18.         self.cat_annotators = 0
  19.         self.cont_annotators = 0
  20.         self.set_imsize(image_size)
  21.         self.set_bbox(person[0])
  22.         self.set_cat(person[1])
  23.         self.set_cont(person[2])
  24.         self.check_cont()
  26.     def set_imsize(self, image_size):
  27.         image_size = np.array(image_size).flatten().tolist()[0]
  28.         row = np.array(image_size[0]).flatten().tolist()[0]
  29.         col = np.array(image_size[1]).flatten().tolist()[0]
  30.         self.im_size.append(row)
  31.         self.im_size.append(col)
  33.     def validate_bbox(self, bbox):
  34.         x1, y1, x2, y2 = bbox
  35.         x1 = min(self.im_size[0], max(0, x1))
  36.         x2 = min(self.im_size[0], max(0, x2))
  37.         y1 = min(self.im_size[1], max(0, y1))
  38.         y2 = min(self.im_size[1], max(0, y2))
  39.         return [int(x1), int(y1), int(x2), int(y2)]
  41.     def set_bbox(self, person_bbox):
  42.         self.bbox = self.validate_bbox(np.array(person_bbox).flatten().tolist())
  44.     def set_cat(self, person_cat):
  45.         cat = np.array(person_cat).flatten().tolist()
  46.         cat = np.array(cat[0]).flatten().tolist()
  47.         self.cat = [np.array(c).flatten().tolist()[0] for c in cat]
  48.         self.cat_annotators = 1
  50.     def set_cont(self, person_cont):
  51.         cont = np.array(person_cont).flatten().tolist()[0]
  52.         self.cont = [np.array(c).flatten().tolist()[0] for c in cont]
  53.         self.cont_annotators = 1
  55.     def check_cont(self):
  56.         for c in self.cont:
  57.             if np.isnan(c):
  58.                 self.cont_annotators = 0
  59.                 break
  61. class emotic_test:
  62.     def __init__(self, filename, folder, image_size, person):
  63.         self.filename = filename
  64.         self.folder = folder
  65.         self.im_size = []
  66.         self.bbox = []
  67.         self.cat = []
  68.         self.cat_annotators = 0
  69.         self.comb_cat = []
  70.         self.cont_annotators = 0
  71.         self.cont = []
  72.         self.comb_cont = []
  73.         self.gender = person[5][0]
  74.         self.age = person[6][0]
  76.         self.set_imsize(image_size)
  77.         self.set_bbox(person[0])
  78.         self.set_cat(person[1])
  79.         self.set_comb_cat(person[2])
  80.         self.set_cont(person[3])
  81.         self.set_comb_cont(person[4])
  82.         self.check_cont()
  84.     def set_imsize(self, image_size):
  85.         image_size = np.array(image_size).flatten().tolist()[0]
  86.         row = np.array(image_size[0]).flatten().tolist()[0]
  87.         col = np.array(image_size[1]).flatten().tolist()[0]
  88.         self.im_size.append(row)
  89.         self.im_size.append(col)
  91.     def validate_bbox(self, bbox):
  92.         x1, y1, x2, y2 = bbox
  93.         x1 = min(self.im_size[0], max(0, x1))
  94.         x2 = min(self.im_size[0], max(0, x2))
  95.         y1 = min(self.im_size[1], max(0, y1))
  96.         y2 = min(self.im_size[1], max(0, y2))
  97.         return [int(x1), int(y1), int(x2), int(y2)]
  99.     def set_bbox(self, person_bbox):
  100.         self.bbox = self.validate_bbox(np.array(person_bbox).flatten().tolist())
  102.     def set_cat(self, person_cat):
  103.         self.cat_annotators = len(person_cat[0])
  104.         for ann in range(self.cat_annotators):
  105.             ann_cat = person_cat[0][ann]
  106.             ann_cat = np.array(ann_cat).flatten().tolist()
  107.             ann_cat = np.array(ann_cat[0]).flatten().tolist()
  108.             ann_cat = [np.array(c).flatten().tolist()[0] for c in ann_cat]
  109.             self.cat.append(ann_cat)
  111.     def set_comb_cat(self, person_comb_cat):
  112.         if self.cat_annotators != 0:
  113.             self.comb_cat = [np.array(c).flatten().tolist()[0] for c in person_comb_cat[0]]
  114.         else:
  115.             self.comb_cat = []
  117.     def set_comb_cont(self, person_comb_cont):
  118.         if self.cont_annotators != 0:
  119.             comb_cont = [np.array(c).flatten().tolist()[0] for c in person_comb_cont[0]]
  120.             self.comb_cont = [np.array(c).flatten().tolist()[0] for c in comb_cont[0]]
  121.         else:
  122.             self.comb_cont = []
  124.     def set_cont(self, person_cont):
  125.         self.cont_annotators = len(person_cont[0])
  126.         for ann in range(self.cont_annotators):
  127.             ann_cont = person_cont[0][ann]
  128.             ann_cont = np.array(ann_cont).flatten().tolist()
  129.             ann_cont = np.array(ann_cont[0]).flatten().tolist()
  130.             ann_cont = [np.array(c).flatten().tolist()[0] for c in ann_cont]
  131.             self.cont.append(ann_cont)
  133.     def check_cont(self):
  134.         for c in self.comb_cont:
  135.             if np.isnan(c):
  136.                 self.cont_annotators = 0
  137.                 break
  139. def cat_to_one_hot(y_cat):
  140.     '''
  141.     One hot encode a categorical label. 
  142.     :param y_cat: Categorical label.
  143.     :return: One hot encoded categorical label. 
  144.     '''
  145.     one_hot_cat = np.zeros(26)
  146.     for em in y_cat:
  147.         one_hot_cat[cat2ind[em]] = 1
  148.     return one_hot_cat
  150. def prepare_data(data_mat, data_path_src, save_dir, dataset_type='train', generate_npy=False, debug_mode=False):
  151.   '''
  152.   Prepare csv files and save preprocessed data in npy files. 
  153.   :param data_mat: Mat data object for a label. 
  154.   :param data_path_src: Path of the parent directory containing the emotic images folders (mscoco, framesdb, emodb_small, ade20k)
  155.   :param save_dir: Path of the directory to save the csv files and the npy files (if generate_npy files is True)
  156.   :param dataset_type: Type of the dataset (train, val or test). Variable used in the name of csv files and npy files. 
  157.   :param generate_npy: If True the data is preprocessed and saved in npy files. Npy files are later used for training. 
  158.   '''
  159.   data_set = list()
  161.   if generate_npy:
  162.     context_arr = list()
  163.     body_arr = list()
  164.     cat_arr = list()
  165.     cont_arr = list()
  166.   
  167.   to_break = 0
  168.   path_not_exist = 0
  169.   cat_cont_zero = 0
  170.   idx = 0
  171.   for ex_idx, ex in enumerate(data_mat[0]):
  172.     nop = len(ex[4][0])
  173.     for person in range(nop):
  174.       if dataset_type == 'train':
  175.         et = emotic_train(ex[0][0],ex[1][0],ex[2],ex[4][0][person])
  176.       else:
  177.         et = emotic_test(ex[0][0],ex[1][0],ex[2],ex[4][0][person])
  178.       try:
  179.         image_path = os.path.join(data_path_src,et.folder,et.filename)
  180.         if not os.path.exists(image_path):
  181.           path_not_exist += 1
  182.           print ('path not existing', ex_idx, image_path)
  183.           continue
  184.         else:
  185.           context = cv2.cvtColor(cv2.imread(image_path),cv2.COLOR_BGR2RGB)
  186.           body = context[et.bbox[1]:et.bbox[3],et.bbox[0]:et.bbox[2]].copy()
  187.           context_cv = cv2.resize(context, (224,224))
  188.           body_cv = cv2.resize(body, (128,128))
  189.       except Exception as e:
  190.         to_break += 1
  191.         if debug_mode == True:
  192.             print ('breaking at idx=%d, %d due to exception=%r' %(ex_idx, idx, e))
  193.         continue
  194.       if (et.cat_annotators == 0 or et.cont_annotators == 0):
  195.         cat_cont_zero += 1
  196.         continue
  197.       data_set.append(et)  
  198.       if generate_npy == True:
  199.         context_arr.append(context_cv)
  200.         body_arr.append(body_cv)
  201.         if dataset_type == 'train':
  202.           cat_arr.append(cat_to_one_hot(et.cat))
  203.           cont_arr.append(np.array(et.cont))
  204.         else: 
  205.           cat_arr.append(cat_to_one_hot(et.comb_cat))
  206.           cont_arr.append(np.array(et.comb_cont))
  207.       if idx % 1000 == 0 and debug_mode==False:
  208.         print (" Preprocessing data. Index = ", idx)
  209.       elif idx % 20 == 0 and debug_mode==True:
  210.         print (" Preprocessing data. Index = ", idx)
  211.       idx = idx + 1
  212.     # for debugging purposes
  213.     if debug_mode == True and idx >= 104:
  214.       print (' ######## Breaking data prep step', idx, ex_idx, ' ######')
  215.       print (to_break, path_not_exist, cat_cont_zero)
  216.       cv2.imwrite(os.path.join(save_dir, 'context1.png'), context_arr[-1])
  217.       cv2.imwrite(os.path.join(save_dir, 'body1.png'), body_arr[-1])
  218.       break
  219.   print (to_break, path_not_exist, cat_cont_zero)
  220.   
  221.   csv_path = os.path.join(save_dir, "%s.csv" %(dataset_type))
  222.   with open(csv_path, 'w') as csvfile:
  223.     filewriter = csv.writer(csvfile, delimiter=',', dialect='excel')
  224.     row = ['Index', 'Folder', 'Filename', 'Image Size', 'BBox', 'Categorical_Labels', 'Continuous_Labels', 'Gender', 'Age']
  225.     filewriter.writerow(row)
  226.     for idx, ex in enumerate(data_set):
  227.         if dataset_type == 'train':
  228.             row = [idx, ex.folder, ex.filename, ex.im_size, ex.bbox, ex.cat, ex.cont, ex.gender, ex.age]
  229.         else:
  230.             row = [idx, ex.folder, ex.filename, ex.im_size, ex.bbox, ex.comb_cat, ex.comb_cont, ex.gender, ex.age]
  231.         filewriter.writerow(row)
  232.   print ('wrote file ', csv_path)
  234.   if generate_npy == True: 
  235.     context_arr = np.array(context_arr)
  236.     body_arr = np.array(body_arr)
  237.     cat_arr = np.array(cat_arr)
  238.     cont_arr = np.array(cont_arr)
  239.     print (len(data_set), context_arr.shape, body_arr.shape)
  240.     np.save(os.path.join(save_dir,'%s_context_arr.npy' %(dataset_type)), context_arr)
  241.     np.save(os.path.join(save_dir,'%s_body_arr.npy' %(dataset_type)), body_arr)
  242.     np.save(os.path.join(save_dir,'%s_cat_arr.npy' %(dataset_type)), cat_arr)
  243.     np.save(os.path.join(save_dir,'%s_cont_arr.npy' %(dataset_type)), cont_arr)
  244.     print (context_arr.shape, body_arr.shape, cat_arr.shape, cont_arr.shape)
  245.   print ('completed generating %s data files' %(dataset_type))
  246.  
  248. def parse_args():
  249.     parser = argparse.ArgumentParser()
  250.     parser.add_argument('--data_dir', type=str,default=r'I:\project\deblur\emotic', help='Path to Emotic data and annotations')
  251.     parser.add_argument('--save_dir_name', type=str, default='emotic_pre', help='Directory name in which preprocessed data will be stored')
  252.     parser.add_argument('--label', type=str,  default='all', choices=['train', 'val', 'test', 'all'])
  253.     parser.add_argument('--generate_npy', action='store_true', help='Generate npy files')
  254.     parser.add_argument('--debug_mode', action='store_true', help='Debug mode. Will only save a small subset of the data')
  255.     # Generate args
  256.     args = parser.parse_args()
  257.     return args
  258.   
  259. if __name__ == '__main__':
  260.     args = parse_args()
  261.     ann_path_src = os.path.join(args.data_dir, 'Annotations','Annotations.mat')
  262.     data_path_src = os.path.join(args.data_dir, 'emotic')
  263.     save_path = os.path.join(args.data_dir, args.save_dir_name)
  264.     if not os.path.exists(save_path):
  265.       os.makedirs(save_path)
  266.     
  267.     cat = ['Affection', 'Anger', 'Annoyance', 'Anticipation', 'Aversion', 'Confidence', 'Disapproval', 'Disconnection',
  268.        'Disquietment', 'Doubt/Confusion', 'Embarrassment', 'Engagement', 'Esteem', 'Excitement', 'Fatigue', 'Fear',
  269.        'Happiness', 'Pain', 'Peace', 'Pleasure', 'Sadness', 'Sensitivity', 'Suffering', 'Surprise', 'Sympathy', 'Yearning']
  270.     cat2ind = {}
  271.     ind2cat = {}
  272.     for idx, emotion in enumerate(cat):
  273.         cat2ind[emotion] = idx
  274.         ind2cat[idx] = emotion
  275.     
  276.     print ('loading Annotations')
  277.     mat = loadmat(ann_path_src)
  278.     if args.label.lower() == 'all':
  279.       labels = ['train', 'val', 'test']
  280.     else:
  281.       labels = [args.label.lower()]
  282.     for label in labels:
  283.       data_mat = mat[label]
  284.       print ('starting label ', label)
  285.       prepare_data(data_mat, data_path_src, save_path, dataset_type=label, generate_npy=args.generate_npy, debug_mode=args.debug_mode)

创作挑战赛

新人创作奖励来咯，坚持创作打卡瓜分现金大奖

版权说明 : 本文为转载文章, 版权归原作者所有 版权申明

原文链接 : https://blog.csdn.net/jacke121/article/details/124209551

内容来源于网络，如有侵权，请联系作者删除！

pythonmatlabpython基础matlab入门

相关文章

    查看更多

    热门标签

    更多
    JavaquerypythonNode开发语言requestUtil数据库Table后端算法LoggerMessageElementParser

    最新文章

    更多
    • 蜀ICP备13028337号-1 大数据知识库 https://www.saoniuhuo.com © All rights reserved
    • 本站内容来源互联网,如果侵犯您的权益请联系我们删除, 联系方式：448109455@qq.com