人脸识别：insightface自定义数据集制作 | 附练手数据集

# 输入nvcc -V# 输出结果nvcc: NVIDIA (R) Cuda compiler driverCopyright (c) 2005-2017 NVIDIA CorporationBuilt on Fri_Sep__1_21:08:03_CDT_2017Cuda compilation tools, release 9.0, V9.0.176

pip install mxnet-cu90# 根据自身情况查询对应的安装命令，例如我服务器的cuda版本是10.0的，安装命令为 pip install mxnet-cu100

git clone --recursive https://github.com/deepinsight/insightface.git

python align_lfw.py --input-dir ../../datasets/lfwdata --output-dir ../../datasets/output

# 重命名 mv [原始文件名] [改变后的文件名] mv lst train.lst# 移动 mv [起始文件路径] [目标文件路径]mv train.lst ../train/

1       ../../datasets/train/Abbas_Kiarostami/Abbas_Kiarostami_0001.jpg 01       ../../datasets/train/Abdel_Aziz_Al-Hakim/Abdel_Aziz_Al-Hakim_0001.jpg   1# 其中1代表对齐，最后的数字0,1代表class label 中间是地址；需要用\t表示tab键，不能用空格间隔。

1000,112,112

python face2rec2.py ../../datasets/train/

s = mx.recordio.pack(header, b'')

train.idx和train.rec

图A的目录 空格 图B的目录 空格 标志0/1（代表两张图类别一致否）

# coding:utf-8import sysimport osimport randomimport timeimport itertoolsimport pdbimport argparse#src = '../../datasets/lfw2'#dst = open('../../datasets/lfw/train.txt', 'a')parser = argparse.ArgumentParser(description='generate image pairs')# generalparser.add_argument('--data-dir', default='', help='')parser.add_argument('--outputtxt', default='', help='path to save.')parser.add_argument('--num-samepairs',default=100)args = parser.parse_args()cnt = 0same_list = []diff_list = []list1 = []list2 = []folders_1 = os.listdir(args.data_dir)dst = open(args.outputtxt, 'a')count = 0dst.writelines('\n')# 产生相同的图像对for folder in folders_1:    sublist = []    same_list = []    imgs = os.listdir(os.path.join(args.data_dir, folder))    for img in imgs:        img_root_path = os.path.join(args.data_dir, folder, img)        sublist.append(img_root_path)        list1.append(img_root_path)    for item in itertools.combinations(sublist, 2):        for name in item:            same_list.append(name)    if len(same_list) > 10 and len(same_list) < 13:        for j in range(0, len(same_list), 2):                if count < int(args.num_samepairs):#数量可以修改                    dst.writelines(same_list[j] + ' ' + same_list[j+1]+ ' ' + '1' + '\n')                    count += 1    if count >= int(args.num_samepairs):        breaklist2 = list1.copy()# 产生不同的图像对diff = 0print(count)# 如果不同的图像对远远小于相同的图像对，则继续重复产生，直到两者相差很小while True:    random.seed(time.time() * 100000 % 10000)    random.shuffle(list2)    for p in range(0, len(list2) - 1, 2):        if list2[p] != list2[p + 1]:            dst.writelines(list2[p] + ' ' + list2[p + 1] + ' ' + '0'+ '\n')            diff += 1            if diff >= count:                break            #print(diff)    if diff < count:        #print('--')        continue    else:        break

python3 generate_image_pairs.py --data-dir ../../datasets/output --outputtxt ../../datasets/train/train.txt --num-samepairs 5

../../datasets/output/Abdullah/Abdullah_0002.jpg ../../datasets/output/Abdullah/Abdullah_0004.jpg 1# 前面两个分布是生成的pairs的路径，后面的0/1代表是否同一个人或类

import mxnet as mxfrom mxnet import ndarray as ndimport argparseimport pickleimport sysimport ossys.path.append(os.path.join(os.path.dirname(__file__), '..', 'eval'))import lfwparser = argparse.ArgumentParser(description='Package LFW images')# generalparser.add_argument('--data-dir', default='', help='')# 修改１：图像大小修改为112，112parser.add_argument('--image-size', type=str, default='112,112', help='')parser.add_argument('--output', default='', help='path to save.')# 修改2：添加解析参数　parser.add_argument('--num-samepairs',default=100)args = parser.parse_args()lfw_dir = args.data_dirimage_size = [int(x) for x in args.image_size.split(',')]# 修改3：将文件名pairs.txt修改成train.txtlfw_pairs = lfw.read_pairs(os.path.join(lfw_dir, 'train.txt'))print(lfw_pairs)# 修改4：下一行进行修改成需要的格式# lfw_paths, issame_list = lfw.get_paths(lfw_dir, lfw_pairs, 'jpg'）lfw_paths, issame_list = lfw.get_paths(lfw_pairs,int(args.num_samepairs)+1)#, 'jpg')lfw_bins = []#lfw_data = nd.empty((len(lfw_paths), 3, image_size[0], image_size[1]))print(len(issame_list))i = 0for path in lfw_paths:  with open(path, 'rb') as fin:    _bin = fin.read()    lfw_bins.append(_bin)    #img = mx.image.imdecode(_bin)    #img = nd.transpose(img, axes=(2, 0, 1))    #lfw_data[i][:] = img    i+=1    if i%1000==0:      print('loading lfw', i)with open(args.output, 'wb') as f:  pickle.dump((lfw_bins, issame_list), f, protocol=pickle.HIGHEST_PROTOCOL)

def get_paths(pairs, same_pairs):    nrof_skipped_pairs = 0    path_list = []    issame_list = []    cnt = 1    for pair in pairs:      path0 = pair[0]      path1 = pair[1]      if cnt < same_pairs:        issame = True      else:        issame = False       if os.path.exists(path0) and os.path.exists(path1):    # Only add the pair if both paths exist        path_list += (path0,path1)        issame_list.append(issame)      else:        print('not exists', path0, path1)        nrof_skipped_pairs += 1      cnt += 1    if nrof_skipped_pairs>0:        print('Skipped %d image pairs' % nrof_skipped_pairs)    return path_list, issame_list

多行注释：1. 进入命令行模式，按ctrl + v进入 visual block模式，然后按j, 或者k选中多行，把需要注释的行标记起来2. 按大写字母I，再插入注释符，例如 #3. 按esc键就会全部注释了取消多行注释：1. 进入命令行模式，按ctrl + v进入 visual block模式，按字母l横向选中列的个数，例如 # 需要选中2列2. 按字母j，或者k选中注释符号3. 按d键就可全部取消注释

python3 lfw2pack.py --data-dir ../../datasets/train --output ../../datasets/train/lfw.bin --num-samepairs 2

    path0 = pair[0]IndexError: list index out of range