使用迁移学习完成图像的多标签分类任务

本文通过迁移学习将预训练好的VGG16模型应用到图像的多标签（Multi-Label）分类问题中。该项目数据来自于Kaggle，每张图片可同时属于多个标签。模型的准确度使用F_{\beta} \text{ score}进行量化，如下表所示：

标签	预测为Positive (1)	预测为Negative (0)
真值为Positive (1)	TP	FN
真值为Negative (0)	FP	TN

例如假设真实标签是(1,0,1,1,0,0)，预测标签是(1,1,0,1,1,0)，则TP=2, FN=1, FP=2, TN=1。Precision=\frac{TP}{TP+FP},\text{  }Recall=\frac{TP}{TP+FN},\text{  }F_{\beta}=\frac{(1+\beta^2)*Presicion*Recall}{Recall+\beta^2*Precision}其中\beta越小，F_{\beta} \text{ score}中Precision的权重越大，\beta等于0时F_{\beta} \text{ score}就变为Precision；\beta越大，F_{\beta} \text{ score}中Recall的权重越大，\beta趋于无穷大时F_{\beta} \text{ score}就变为Recall。可以在Keras中自定义该函数（y_pred表示预测概率）：

此外在损失函数的使用上多标签分类和多类别（multi-class）分类也有区别，多标签分类使用\text{binary crossentropy loss}，假设一个样本的真实标签是(1,0,1,1,0,0)，预测概率是(0.2, 0.3, 0.4, 0.7, 0.9, 0.2)： \text{binary crossentropy loss}=-(\ln 0.2 + \ln 0.7 + \ln 0.4 + \ln 0.7 + \ln 0.1 + \ln 0.8)/6=0.96另外多标签分类输出层的激活函数应选择sigmoid而非softmax。模型架构如下所示：

1. 从Kaggle网站上下载数据并解压，将其处理成可被模型读取的数据格式：

点击查看代码

from os import listdir from numpy import zeros, asarray, savez_compressed from pandas import read_csv from tensorflow.keras.preprocessing.image import load_img, img_to_array # create a mapping of tags to integers given the loaded mapping file def create_tag_mapping(mapping_csv): labels = set() # create a set of all known tags for i in range(len(mapping_csv)): tags = mapping_csv['tags'][i].split(' ') # convert spaced separated tags into an array of tags labels.update(tags) # add tags to the set of known labels labels = sorted(list(labels)) # convert set of labels to a sorted list # dict that maps labels to integers, and the reverse labels_map = {labels[i]:i for i in range(len(labels))} inv_labels_map = {i:labels[i] for i in range(len(labels))} return labels_map, inv_labels_map # create a mapping of filename to a list of tags def create_file_mapping(mapping_csv): mapping = dict() for i in range(len(mapping_csv)): name, tags = mapping_csv['image_name'][i], mapping_csv['tags'][i] mapping[name] = tags.split(' ') return mapping # create a one hot encoding for one list of tags def one_hot_encode(tags, mapping): encoding = zeros(len(mapping), dtype='uint8') # create empty vector # mark 1 for each tag in the vector for tag in tags: encoding[mapping[tag]] = 1 return encoding # load all images into memory def load_dataset(path, file_mapping, tag_mapping): photos, targets = list(), list() # enumerate files in the directory for filename in listdir(path): photo = load_img(path + filename, target_size=(128,128)) # load image photo = img_to_array(photo, dtype='uint8') # convert to numpy array tags = file_mapping[filename[:-4]] # get tags target = one_hot_encode(tags, tag_mapping) # one hot encode tags photos.append(photo) targets.append(target) X = asarray(photos, dtype='uint8') y = asarray(targets, dtype='uint8') return X, y filename = 'train_v2.csv' # load the target file mapping_csv = read_csv(filename) tag_mapping, _ = create_tag_mapping(mapping_csv) # create a mapping of tags to integers file_mapping = create_file_mapping(mapping_csv) # create a mapping of filenames to tag lists folder = 'train-jpg/' # load the jpeg images X, y = load_dataset(folder, file_mapping, tag_mapping) print(X.shape, y.shape) savez_compressed('planet_data.npz', X, y) # save both arrays to one file in compressed format

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

from os import listdir from numpy import zeros, asarray, savez_compressed from pandas import read_csv from tensorflow.keras.preprocessing.image import load_img, img_to_array   # create a mapping of tags to integers given the loaded mapping file def create_tag_mapping(mapping_csv):     labels = set() # create a set of all known tags     for i in range(len(mapping_csv)):         tags = mapping_csv['tags'][i].split(' ') # convert spaced separated tags into an array of tags         labels.update(tags) # add tags to the set of known labels     labels = sorted(list(labels)) # convert set of labels to a sorted list     # dict that maps labels to integers, and the reverse     labels_map = {labels[i]:i for i in range(len(labels))}     inv_labels_map = {i:labels[i] for i in range(len(labels))}     return labels_map, inv_labels_map   # create a mapping of filename to a list of tags def create_file_mapping(mapping_csv):     mapping = dict()     for i in range(len(mapping_csv)):         name, tags = mapping_csv['image_name'][i], mapping_csv['tags'][i]         mapping[name] = tags.split(' ')     return mapping   # create a one hot encoding for one list of tags def one_hot_encode(tags, mapping):     encoding = zeros(len(mapping), dtype='uint8') # create empty vector     # mark 1 for each tag in the vector     for tag in tags: encoding[mapping[tag]] = 1     return encoding   # load all images into memory def load_dataset(path, file_mapping, tag_mapping):     photos, targets = list(), list()     # enumerate files in the directory     for filename in listdir(path):         photo = load_img(path + filename, target_size=(128,128)) # load image         photo = img_to_array(photo, dtype='uint8') # convert to numpy array         tags = file_mapping[filename[:-4]] # get tags         target = one_hot_encode(tags, tag_mapping) # one hot encode tags         photos.append(photo)         targets.append(target)     X = asarray(photos, dtype='uint8')     y = asarray(targets, dtype='uint8')     return X, y   filename = 'train_v2.csv' # load the target file mapping_csv = read_csv(filename) tag_mapping, _ = create_tag_mapping(mapping_csv) # create a mapping of tags to integers file_mapping = create_file_mapping(mapping_csv) # create a mapping of filenames to tag lists folder = 'train-jpg/' # load the jpeg images X, y = load_dataset(folder, file_mapping, tag_mapping) print(X.shape, y.shape) savez_compressed('planet_data.npz', X, y) # save both arrays to one file in compressed format

2. 建立两个辅助函数，第一个函数用来分割训练集和验证集，第二个函数用来画出模型在训练过程中的学习曲线：

3. 使用Data Augmentation增加样本并对模型进行训练：