응급 상황 분류

from keras.models import Model from keras.layers import Input, Dense, LSTM, Dropout, Reshape, Conv2D, MaxPooling2D, Flatten, concatenate from transformers import TFBertModel # Define input shapes text_input_shape = (None,) image_input_shape = (128, 128, 3) # Define the text model using Kobert text_input = Input(shape=text_input_shape, dtype='int32') bert_model = TFBertModel.from_pretrained('monologg/kobert') text_output = bert_model(text_input)[1] # Define the image model using convolutional layers and LSTM image_input = Input(shape=image_input_shape) conv1 = Conv2D(32, kernel_size=(3, 3), activation='relu')(image_input) pool1 = MaxPooling2D(pool_size=(2, 2))(conv1) conv2 = Conv2D(64, kernel_size=(3, 3), activation='relu')(pool1) pool2 = MaxPooling2D(pool_size=(2, 2))(conv2) conv3 = Conv2D(128, kernel_size=(3, 3), activation='relu')(pool2) reshape = Reshape((-1, 128))(conv3) lstm = LSTM(64, return_sequences=True)(reshape) dropout = Dropout(0.5)(lstm) flatten = Flatten()(dropout) # Combine the text and image models combined = concatenate([text_output, flatten]) # Add a dense layer with 128 units and ReLU activation dense1 = Dense(128, activation='relu')(combined) # Add a dropout layer to prevent overfitting dropout2 = Dropout(0.5)(dense1) # Add a dense layer with 7 units and softmax activation for classification output = Dense(7, activation='softmax')(dropout2) # Define the model model = Model(inputs=[text_input, image_input], outputs=output) # Compile the model model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])