生成BiFPN结构热力图的代码

下面是使用Python和TensorFlow实现生成BiFPN结构热力图的代码示例：

import tensorflow as tf
from tensorflow.keras.layers import Input, Conv2D, MaxPooling2D, UpSampling2D, concatenate, BatchNormalization, Activation
from tensorflow.keras.models import Model
from tensorflow.keras.utils import plot_model
import numpy as np

def build_bifpn(input_shape, num_channels, num_layers, num_filters):
    # 定义输入层
    inputs = Input(input_shape)
    # 定义第一层
    conv1 = Conv2D(num_filters, (3,3), padding='same', activation='relu')(inputs)
    pool1 = MaxPooling2D(pool_size=(2,2))(conv1)
    # 定义多层bifpn
    layers = [pool1]
    for i in range(num_layers):
        # 前向传播
        conv = Conv2D(num_filters, (3,3), padding='same', activation='relu')(layers[-1])
        up = UpSampling2D(size=(2,2))(conv)
        merge = concatenate([layers[-2], up], axis=3)
        conv1 = Conv2D(num_filters, (3,3), padding='same', activation='relu')(merge)
        conv2 = Conv2D(num_filters, (1,1), padding='same')(conv1)
        bn = BatchNormalization()(conv2)
        activation = Activation('relu')(bn)
        # 添加到层列表中
        layers.append(activation)
    # 定义输出层
    output = layers[-1]
    output = Conv2D(num_channels, (1,1), padding='same', activation='sigmoid')(output)
    # 构建模型
    model = Model(inputs=[inputs], outputs=[output])
    return model

# 定义输入形状，通道数和层数
input_shape = (128, 128, 3)
num_channels = 1
num_layers = 3
num_filters = 64

# 构建模型
model = build_bifpn(input_shape, num_channels, num_layers, num_filters)

# 绘制模型结构图
plot_model(model, to_file='bifpn.png', show_shapes=True)

# 生成热力图
from tensorflow.keras.utils import plot_model
import matplotlib.pyplot as plt

layer_names = []
for layer in model.layers:
    layer_names.append(layer.name)

activations = model.predict(np.random.rand(1, 128, 128, 3))

images_per_row = 16

for layer_name, layer_activation in zip(layer_names, activations):
    n_features = layer_activation.shape[-1]
    size = layer_activation.shape[1]
    n_cols = n_features // images_per_row
    display_grid = np.zeros((size * n_cols, images_per_row * size))
    for col in range(n_cols):
        for row in range(images_per_row):
            channel_image = layer_activation[0, :, :, col * images_per_row + row]
            channel_image -= channel_image.mean()
            channel_image /= channel_image.std()
            channel_image *= 64
            channel_image += 128
            channel_image = np.clip(channel_image, 0, 255).astype('uint8')
            display_grid[col * size : (col + 1) * size, row * size : (row + 1) * size] = channel_image
    scale = 1. / size
    plt.figure(figsize=(scale * display_grid.shape[1], scale * display_grid.shape[0]))
    plt.title(layer_name)
    plt.grid(False)
    plt.imshow(display_grid, aspect='auto', cmap='viridis')
    
plt.show()

上述代码中使用了TensorFlow的keras API来构建BiFPN模型，并通过plot_model函数将模型结构绘制成图形化的png文件。使用模型的predict函数获取模型的输出，然后通过循环遍历每一层的输出，将其可视化成热力图。具体来说，将每一层的输出展开成网格，并将每个通道的特征图在网格上显示，同时对通道进行标准化和调整亮度，以便更好地观察特征图的细节。最后，将所有热力图显示在同一张图中，以便比较不同层之间的特征变化。