keras - 召回率和精度 0.00e+00

Question

我正在用 1280 个大脑 MIP 图像训练模型以检测脑动脉瘤：

    # 1*1 convolutional layers
    
    conv_final = layers.Conv2D(NUM_CLASSES, kernel_size=(1,1))(up_conv_512)
    conv_final = layers.BatchNormalization(axis=axis)(conv_final)
    conv_final = layers.Activation('sigmoid')(conv_final)  #Change to softmax for multichannel

    # Model integration
    model = models.Model(inputs, conv_final, name="AttentionResUNet")
    return model
input_shape = (512,512,3)
model=Attention_ResUNet(input_shape, NUM_CLASSES=1, dropout_rate=0.2, batch_norm=True)
model.summary()

为了训练：

model =  Attention_ResUNet(input_shape)
    metrics = [ dice_coef,Recall(), Precision()]
    model.compile(loss=dice_coef_loss, optimizer=Adam(lr), metrics=metrics)

几个 epoch 的召回率和精度为 0.00e+00 可能有什么问题？

标签： kerasmodelprecision-recall