基于CNN的人脸微笑检测模型训练报错：数组真值判断歧义

问题根源 #

错误出在class_weight参数的格式上：代码中计算的classWeight是numpy数组，但Keras的model.fit()要求class_weight为字典类型（键为类别索引，值为对应权重）。传入数组时，Keras内部尝试判断其真值，触发了"数组真值不明确"的报错。此外代码还存在拼写错误、导入缺失和缩进问题，也需要同步修正。

修正步骤 #

1. 补全缺失导入并修正拼写错误 #

添加必要库的导入，同时修正rom imutils import paths的拼写错误：

import numpy as np
import os
import imutils
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, Activation, MaxPooling2D, Flatten, Dense
from tensorflow.keras.utils import np_utils

2. 转换class_weight为字典格式 #

将原数组格式的权重转换为Keras要求的字典结构：

classTotals = labels.sum(axis=0)
# 生成{类别索引: 权重值}的字典
classWeight = {i: classTotals.max() / classTotals[i] for i in range(len(classTotals))}

3. 修正LeNet类的缩进问题 #

原代码中build方法内的层添加代码缩进错误，导致方法返回空模型。修正后加上@staticmethod装饰器，确保无需实例化即可调用：

class LeNet:
    @staticmethod
    def build(width, height, depth, classes):
        model = Sequential()
        inputShape = (height, width, depth)

        # 第一层CONV=>RELU=>POOL
        model.add(Conv2D(20, (5, 5), padding="same", input_shape=inputShape))
        model.add(Activation("relu"))
        model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

        # 第二层CONV=>RELU=>POOL
        model.add(Conv2D(50, (5, 5), padding="same"))
        model.add(Activation("relu"))
        model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

        # 全连接层与扁平化
        model.add(Flatten())
        model.add(Dense(500))
        model.add(Activation("relu"))

        model.add(Dense(classes))
        model.add(Activation("softmax"))
        return model

完整修正后的代码 #

from google.colab import drive
drive.mount('/content/drive')

# 补全缺失导入
import numpy as np
import os
import imutils
from imutils import paths
import cv2
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, Activation, MaxPooling2D, Flatten, Dense
from tensorflow.keras.utils import img_to_array, np_utils

data = []
labels = []

imagePaths = list(paths.list_images("/content/drive/MyDrive/dataset/SMILEs"))

for imagePath in sorted(imagePaths):
    # 图像预处理
    image = cv2.imread(imagePath)
    image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    image = imutils.resize(image, width=28)
    image = img_to_array(image)
    data.append(image)
    # 提取标签
    label = imagePath.split(os.path.sep)[-3]
    label = "smiling" if label == "positives" else "not_smiling"
    labels.append(label)

# 数据标准化与标签编码
data = np.array(data, dtype="float") / 255.0
labels = np.array(labels)

le = LabelEncoder().fit(labels)
labels = np_utils.to_categorical(le.transform(labels), 2)

# 计算类别权重并转换为字典格式
classTotals = labels.sum(axis=0)
classWeight = {i: classTotals.max() / classTotals[i] for i in range(len(classTotals))}

# 划分训练测试集
(trainX, testX, trainY, testY) = train_test_split(data,labels, test_size=0.20, stratify=labels, random_state=42)

# 定义LeNet模型
class LeNet:
    @staticmethod
    def build(width, height, depth, classes):
        model = Sequential()
        inputShape = (height, width, depth)

        model.add(Conv2D(20, (5, 5), padding="same", input_shape=inputShape))
        model.add(Activation("relu"))
        model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

        model.add(Conv2D(50, (5, 5), padding="same"))
        model.add(Activation("relu"))
        model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))

        model.add(Flatten())
        model.add(Dense(500))
        model.add(Activation("relu"))

        model.add(Dense(classes))
        model.add(Activation("softmax"))
        return model

# 构建并训练模型
model = LeNet.build(28, 28, 1, 2)

print("[INFO] compiling model...")
model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=['accuracy'])

print("[INFO] training network...")
H = model.fit(trainX, trainY, validation_data=(testX, testY), class_weight=classWeight, batch_size=64, epochs=15, verbose=1)

内容的提问来源于stack exchange，提问作者Ibrahim Oussama benhabi