def infer(data, rescale=RESCALE, resize_factor=RESIZE): ## test mode
##### DO NOT CHANGE ORDER OF TEST DATA #####
X = []
for i, d in enumerate(data):
X.append(image_preprocessing(d, rescale, resize_factor))
X = np.array(X)
p1 = m1.predict(X)
X2 = []
for i, d in enumerate(data):
X2.append(image_preprocessing2(d, rescale, resize_factor))
X2 = np.array(X2)
p2 = m2.predict(X2)
# X3 = []
# for i, d in enumerate(data):
# X3.append(image_preprocessing3(d, rescale, resize_factor))
# X3 = np.array(X3)
# p3 = m3.predict(X3)
# X4 = []
# for i, d in enumerate(data):
# X4.append(image_preprocessing4(d, rescale, resize_factor))
# X4 = np.array(X4)
# p4 = m4.predict(X4)
pp1 = []
for p in p1:
p = new_softmax(p)
pp1.append(p)
pp2 = []
for p in p2:
p = new_softmax(p)
pp2.append(p)
# pp3 = []
# for p in p3:
# p = new_softmax(p)
# pp3.append(p)
# pp4 = []
# for p in p4:
# p = new_softmax(p)
# pp4.append(p)
pp1 = np.array(pp1)
pp2 = np.array(pp2)
# pp3 = np.array(pp3)
# pp4 = np.array(pp4)
# X = (pp1 + pp2 + pp3 + pp4) / 4
X = (pp1 + pp2) / 2
pred = np.argmax(X, axis=-1)
print('Prediction done!\n Saving the result...')
return pred
def infer(data, rescale=RESCALE, resize_factor=RESIZE): ## test mode
##### DO NOT CHANGE ORDER OF TEST DATA #####
X = []
for i, d in enumerate(data):
X.append(image_preprocessing(d, rescale, resize_factor))
X = np.array(X)
pred = model.predict(X)
print('Prediction done!\n Saving the result...')
return pred
def infer(data, rescale=RESCALE, resize_factor=RESIZE): ## test mode
##### DO NOT CHANGE ORDER OF TEST DATA #####
X = []
for i, d in enumerate(data):
# test 데이터를 training 데이터와 같이 전처리 하기
X.append(image_preprocessing(d, rescale, resize_factor))
X = np.array(X)
pred = model.predict(X) # 모델 예측 결과: 0-3
pred = np.argmax(pred, axis=1)
print('Prediction done!\n Saving the result...')
return pred
def infer(
data,
target_resolution=TARGET_RESOLUTION,
): ## test mode
##### DO NOT CHANGE ORDER OF TEST DATA #####
X = []
for i, d in enumerate(data):
# test 데이터를 training 데이터와 같이 전처리 하기
X.append(image_preprocessing(d, target_resolution, normalize=True))
X = np.array(X)
pred = model.predict(X)
pred = np.argmax(pred, axis=1) # 모델 예측 결과: 0-3
print('Prediction done!\n Saving the result...')
return pred
def infer(data, rescale=RESCALE, resize_factor=RESIZE): ## test mode
##### DO NOT CHANGE ORDER OF TEST DATA #####
X = []
for i, d in enumerate(data):
# test 데이터를 training 데이터와 같이 전처리 하기
X.append(image_preprocessing(d, rescale, resize_factor))
X = np.array(X)
inception_pred = inception_model.predict(X)
efficient_pred = efficient_model.predict(X)
# mobilenet_pred = mobilenet_model.predict(X)
# resnet_pred = resnet_model.predict(X)
# densenet_pred = densenet_model.predict(X)
pred = (inception_pred * inception_ratio +
efficient_pred * efficient_ratio)
# pred += (mobilenet_pred * mobilenet_ratio)
# pred += (resnet_pred * resnet_ratio + densenet_pred * densenet_ratio)
pred = np.argmax(pred, axis=1)
print('Prediction done!\n Saving the result...')
return pred