Fisher 向量特征编码

Fisher 向量是一种图像特征编码和量化技术，可以看作是流行的视觉词袋或 VLAD 算法的软或概率版本。图像使用视觉词汇建模，该词汇使用在低级图像特征（如 SIFT 或 ORB 描述符）上训练的 K-mode 高斯混合模型进行估计。Fisher 向量本身是高斯混合模型 (GMM) 相对于其参数（混合权重、均值和协方差矩阵）的梯度的串联。

在此示例中，我们为 scikit-learn 中的数字数据集计算 Fisher 向量，并在这些表示上训练分类器。

请注意，运行此示例需要 scikit-learn。

              precision    recall  f1-score   support

           0       0.79      0.85      0.82        48
           1       0.73      0.68      0.70        56
           2       0.60      0.77      0.68        35
           3       0.56      0.57      0.56        44
           4       0.79      0.62      0.70        48
           5       0.56      0.60      0.58        47
           6       0.50      0.63      0.56        38
           7       0.70      0.68      0.69        44
           8       0.65      0.52      0.58        46
           9       0.57      0.52      0.55        44

    accuracy                           0.64       450
   macro avg       0.64      0.65      0.64       450
weighted avg       0.65      0.64      0.64       450

from matplotlib import pyplot as plt
import numpy as np
from sklearn.datasets import load_digits
from sklearn.metrics import classification_report, ConfusionMatrixDisplay
from sklearn.model_selection import train_test_split
from sklearn.svm import LinearSVC

from skimage.transform import resize
from skimage.feature import fisher_vector, ORB, learn_gmm


data = load_digits()
images = data.images
targets = data.target

# Resize images so that ORB detects interest points for all images
images = np.array([resize(image, (80, 80)) for image in images])

# Compute ORB descriptors for each image
descriptors = []
for image in images:
    detector_extractor = ORB(n_keypoints=5, harris_k=0.01)
    detector_extractor.detect_and_extract(image)
    descriptors.append(detector_extractor.descriptors.astype('float32'))

# Split the data into training and testing subsets
train_descriptors, test_descriptors, train_targets, test_targets = train_test_split(
    descriptors, targets
)

# Train a K-mode GMM
k = 16
gmm = learn_gmm(train_descriptors, n_modes=k)

# Compute the Fisher vectors
training_fvs = np.array(
    [fisher_vector(descriptor_mat, gmm) for descriptor_mat in train_descriptors]
)

testing_fvs = np.array(
    [fisher_vector(descriptor_mat, gmm) for descriptor_mat in test_descriptors]
)

svm = LinearSVC().fit(training_fvs, train_targets)

predictions = svm.predict(testing_fvs)

print(classification_report(test_targets, predictions))

ConfusionMatrixDisplay.from_estimator(
    svm,
    testing_fvs,
    test_targets,
    cmap=plt.cm.Blues,
)

plt.show()