在尝试为我的自定义模型生成Grad-CAM时，我遇到了一个问题。我正在尝试使用resnet50对图像分类模型进行微调。我的模型是这样定义的：
IMG_SHAPE = (img_height,img_width) + (3,)base_model = tf.keras.applications.ResNet50(input_shape=IMG_SHAPE, include_top=False, weights='imagenet')然后，
preprocess_input = tf.keras.applications.resnet50.preprocess_input最后，
input_layer = tf.keras.Input(shape=(img_height, img_width, 3),name="input_layer")x = preprocess_input(input_layer)x = base_model(x, training=False)x = tf.keras.layers.GlobalAveragePooling2D(name="global_average_layer")(x)x = tf.keras.layers.Dropout(0.2,name="dropout_layer")(x)x = tf.keras.layers.Dense(4,name="training_layer")(x)outputs = tf.keras.layers.Dense(4,name="prediction_layer")(x)model = tf.keras.Model(input_layer, outputs)现在，我按照https://keras.io/examples/vision/grad_cam/上的教程来生成热力图。但是，尽管教程建议使用model.summary()来获取最后的卷积层和分类层，我不确定如何在我的模型中实现。如果我运行model.summary()，我得到的是：
__________________________________________________________________________________________________Layer (type) Output Shape Param # Connected to ==================================================================================================input_layer (InputLayer) [(None, 224, 224, 3)] 0 __________________________________________________________________________________________________tf.operators.getitem_11 (None, 224, 224, 3) 0 __________________________________________________________________________________________________tf.nn.bias_add_11 (TFOpLambd [(None, 224, 224, 3)] 0__________________________________________________________________________________________________resnet50 (Functional) (None, 7, 7, 2048) 23587712__________________________________________________________________________________________________global_average (GlobalAverag (None, 2048) 0__________________________________________________________________________________________________dropout_layer (Dropout) (None, 2048) 0__________________________________________________________________________________________________hidden_layer (Dense) (None, 4) 8196__________________________________________________________________________________________________predict_layer (Dense) (None, 4) 20==================================================================================================然而，如果我运行base_model.summary()，我得到的是：
Layer (type) Output Shape Param # Connected to ==================================================================================================input_29 (InputLayer) [(None, 224, 224, 3) 0 __________________________________________________________________________________________________conv1_pad (ZeroPadding2D) (None, 230, 230, 3) 0 input_29[0][0] __________________________________________________________________________________________________conv1_conv (Conv2D) (None, 112, 112, 64) 9472 conv1_pad[0][0] __________________________________________________________________________________________________conv1_bn (BatchNormalization) (None, 112, 112, 64) 256 conv1_conv[0][0] __________________________________________________________________________________________________... ... ... ... __________________________________________________________________________________________________conv5_block3_3_bn (BatchNormali (None, 7, 7, 2048) 8192 conv5_block3_3_conv[0][0] __________________________________________________________________________________________________conv5_block3_add (Add) (None, 7, 7, 2048) 0 conv5_block2_out[0][0] conv5_block3_3_bn[0][0] __________________________________________________________________________________________________conv5_block3_out (Activation) (None, 7, 7, 2048) 0 conv5_block3_add[0][0] ==================================================================================================如果我按照教程使用’resnet50’作为最后的卷积层，我会得到以下错误：
图形断开连接：无法获取层“conv1_pad”处张量KerasTensor(type_spec=TensorSpec(shape=(None, 224, 224, 3), dtype=tf.float32, name=’input_29′), name=’input_29′, description=”created by layer ‘input_29′”)的值。之前访问的层没有问题：[]
但是，如果我使用’conv5_block3_out’，程序无法在模型中找到该层。我该如何访问似乎隐藏在resnet50层中的层呢？
回答：
我设法找到了解决这个问题的方法。在定义“make-gradcam_heatmap”时，我添加了以下一行代码
input_layer = model.get_layer('resnet50').get_layer('input_1').input并将下一行改为
last_conv_layer = model.get_layer(last_conv_layer_name).get_layer("conv5_block3_out")