问题描述
我正在编写 DCNN 模型的分类报告,但遇到错误。我的代码是
from sklearn.metrics import confusion_matrix
test = ImageDataGenerator()
test_generator = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255)
test_data = test_generator.flow_from_directory(directory="/content/dataset/test",target_size=IMAGE_SHAPE,color_mode="rgb",class_mode='categorical',batch_size=1,shuffle = False )
test_data.reset()
predicted_class_indices=np.argmax(pred,axis=1)
cm = confusion_matrix(test_labels,predictions.argmax(axis=1))
错误:
AttributeError: 'list' object has no attribute 'argmax'
解决方法
你的 predictions
显然是一个 Python 列表,列表没有 argmax
属性;您需要使用 Numpy 函数 argmax()
:
predictions = [[0.1,0.9],[0.8,0.2]] # dummy data
y_pred_binary = predictions.argmax(axis=1)
# AttributeError: 'list' object has no attribute 'argmax'
# Use Numpy:
import numpy as np
y_pred_binary = np.argmax(predictions,axis=1)
y_pred_binary
# array([1,0])