如何计算测试数据中每张图像的损失和准确性？

Question

我想计算测试数据中每个图像的损失和准确性。这样我就可以计算出平均损失和准确性及其标准偏差。

我正在使用U-Net模型进行图像分割。我将数据分为训练数据和测试数据，它们都有图像和标签。

训练阶段

data_gen_args = dict(rotation_range=0.2, width_shift_range=0.05, height_shift_range=0.05, shear_range=0.05,
                    zoom_range=0.05, horizontal_flip=True, fill_mode='nearest')

trainGene = trainGenerator(1,'./data/melanoma/train','image','label',data_gen_args,save_to_dir = None)

model = unet()

tensorboard = TensorBoard(log_dir='./logs', histogram_freq=0,write_graph=True, write_images=False)

model_checkpoint = ModelCheckpoint('./models/unet_model.hdf5', monitor='loss',verbose=1, save_best_only=True)

history = model.fit_generator(trainGene,steps_per_epoch=2,epochs=1,callbacks=[model_checkpoint,tensorboard])

print(history.history)

print("number of epoch",history.epoch)

有效阶段

data_gen_args = dict(rotation_range=0.2, width_shift_range=0.05, height_shift_range=0.05, shear_range=0.05,
                    zoom_range=0.05, horizontal_flip=True, fill_mode='nearest')

evaluateGene = validGenerator(1,'./data/melanoma/test','image','label',data_gen_args,save_to_dir = None)

score = model.evaluate_generator(evaluateGene,20, verbose=1)

print(model.metrics_names)

print(score[0], score[1],score[2])

预测阶段

testGene = testGenerator("data/melanoma/test/image")

results = model.predict_generator(testGene,20,verbose=1)
saveResult("data/melanoma/test/predict",results)

我希望输出可以是每个测试图像的损失和准确性的数组。 谢谢！