def read_img(img_path):
try:
img = image.load_img(img_path, target_size=(224, 224))
except Exception as e:
print(e)
img = image.img_to_array(img)
img = np.expand_dims(img, axis=0)
# print("loadimg……")
return img / 255
def convert_to_HDF5(dir, data):
for i in np.arange(0, len(dir)):
image = load_img(ROOT_DIR + dir[i], target_size=(224, 224))
image = img_to_array(image)
image = image/255
image = np.expand_dims(image, axis=0)
print(i)
feature, _, __ = model.predict_on_batch([image, image, image])
data[i] = feature
def read_and_prep_images(img_paths,
img_height=image_size,
img_width=image_size):
imgs = [
load_img(img_path, target_size=(img_height, img_width))
for img_path in img_paths
]
img_array = np.array([img_to_array(img) for img in imgs])
return preprocess_input(img_array)
def get_lab_from_data_list(data_list):
x_lab = []
for f in data_list:
rgb = img_to_array(load_img(f,
target_size=(img_size,
img_size))).astype(np.uint8)
lab = rgb2lab(rgb)
x_lab.append(lab)
return np.stack(x_lab)
def predict(path, filename):
x = load_img(path, target_size=(width, height))
x = img_to_array(x)
x = np.expand_dims(x, axis=0)
array = cnn.predict(x)
result = array[0]
answer = np.argmax(result)
return answer
def categorize(self, lmodel, imgName):
#BGR
img = image.load_img(imgName, target_size=(299, 299))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
y = lmodel.predict(x) #
return y #np.argmax(y)
def read_and_prep_images(img_paths):
'''read and prepare images'''
image_size = 224
imgs = [
load_img(img_path, target_size=(image_size, image_size))
for img_path in img_paths
]
img_array = np.array([img_to_array(img) for img in imgs])
return preprocess_input(img_array)
def create_test_data(self):
sar_filename, lab_filename = self.load_sample(
sar_dir='/emwusr/xianzhengshi/2020_test/20200404/test_256/sar',
lab_dir='/emwusr/xianzhengshi/2020_test/20200404/test_256/lab')
test_data = np.ndarray(
(len(sar_filename), self.img_cols, self.img_rows, 3))
test_lab = np.ndarray(
(len(lab_filename), self.img_cols, self.img_rows, self.num_class))
for num, name in enumerate(sar_filename):
img = load_img(name)
img = img_to_array(img)
test_data[num] = img
# test_data[num] = (img[:, :, 0]).reshape((self.img_rows, self.img_cols, 1))
for num1, name1 in enumerate(lab_filename):
lab = load_img(name1)
lab = img_to_array(lab)
test_lab[num1] = self.one_hot_lab(lab)
return test_data, test_lab
def get_batch_x(self, idx):
# Fetch a batch of training data
batch_imgs = self.image_list[idx * BATCH_SIZE:(idx + 1) * BATCH_SIZE]
X = np.array([
img_to_array(load_img(im, target_size=(H, W))) / IMG_NORMALIZER
for im in batch_imgs
])
if PREPROCESS: X = F_PREPROC(X)
return X
def predict(img_local_path):
model = SqueezeNet(weights='imagenet')
img = image.load_img(img_local_path, target_size=(227, 227))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
preds = model.predict(x)
result = decode_predictions(preds)
return result
def run(self, path_image, model='VGG', k=5, load_model=True, load_features=True,
fit_model=True, data_augmentation=False, algorithm='brute', metric='cosine',
nb_imgs=100, remove_not_white=False):
self.path_to_img = path_image
#load the model
if load_model:
self._load_model(model=model)
#load the features
if load_features:
self._load_features(model=model,
data_augmentation=data_augmentation,
remove_not_white=remove_not_white)
#fit the kNN model
if fit_model:
self._fit_kNN(algorithm=algorithm, metric=metric)
#calculate the features of the images
if model=='Inception_Resnet':
img = image.load_img(path_image, target_size=(299, 299))
img = image.img_to_array(img) # convert to array
# if remove_background:
# img = utils.remove_background(img.astype(np.uint8))
img = np.expand_dims(img, axis=0)
img = ppIR(img)
self.img_features = [self.IR_model.predict(img).flatten()]
else:
img = image.load_img(path_image, target_size=(224, 224))
img = image.img_to_array(img) # convert to array
# if remove_background:
# img = utils.remove_background(img.astype(np.uint8))
img = np.expand_dims(img, axis=0)
img = ppVGG19(img)
self.img_features = [self.VGG_model.predict(img).flatten()]
#find most similar images in the dataset
_, self.NN = self.kNN.kneighbors(self.img_features)
print(self.kNN.kneighbors(self.img_features))
#identify most similar items
self.similar_items = [self.items[i] for i in self.NN[0]][:nb_imgs]
self.similar_images = [self.images[i] for i in self.NN[0]][:nb_imgs]
def learn(self, style_name):
img_paths = glob.glob("transfer/ns_model/train_img/*")
batch_size = 2
epochs = 5
input_shape = (224, 224, 3)
input_size = input_shape[:2]
style = glob.glob("media/style/*")[0].split("\\")[-1]
img_sty = load_img(
'media/style/'+style,
target_size=input_size
)
img_arr_sty = np.expand_dims(img_to_array(img_sty), axis=0)
self.y_true_sty = self.model_sty.predict(img_arr_sty)
shutil.rmtree("./media/style")
os.mkdir("./media/style")
self.gen = self.train_generator(
img_paths,
batch_size,
self.model_con,
self.y_true_sty,
epochs=epochs
)
gen_output_layer = self.model_gen.layers[-1]
tv_loss = self.TVRegularizer(gen_output_layer.output)
gen_output_layer.add_loss(tv_loss)
self.model.compile(
optimizer = Adadelta(),
loss = [
self.style_loss,
self.style_loss,
self.style_loss,
self.style_loss,
self.feature_loss
],
loss_weights = [1.0, 1.0, 1.0, 1.0, 3.0]
)
now_epoch = 0
min_loss = np.inf
steps_per_epoch = math.ceil(len(img_paths)/batch_size)
# 学習
for i , (X_train, y_train) in enumerate(self.gen):
if i % steps_per_epoch == 0:
now_epoch += 1
loss = self.model.train_on_batch(X_train, y_train)
if loss[0]<min_loss:
min_loss = loss[0]
self.model.save("transfer/ns_model/pretrained_model/" + style_name + ".h5")
print("epoch: {}, iters: {}, loss: {:.3f}".format(now_epoch, i, loss[0]))
def example():
image_path = 'D:\\Onepredict_MK\\LG CNS\\cat.jpg'
img = image.load_img(image_path, target_size=(224, 224))
img_input = preprocess_input(np.expand_dims(img, 0))
resnet = ResNet50(input_shape=(224, 224, 3),
weights='imagenet',
include_top=True)
probs = resnet.predict(img_input)
pred = np.argmax(probs[0])
activation_layer = resnet.layers[-3].name
inp = resnet.input
# for idx in range(1000):
y_c = resnet.output.op.inputs[0][:, pred]
A_k = resnet.get_layer(activation_layer).output
grads = K.gradients(y_c, A_k)[0]
# Model(inputs=[inp], outputs=[A_k, grads, resnet.output])
get_output = K.function(inputs=[inp], outputs=[A_k, grads, resnet.output])
[conv_output, grad_val, model_output] = get_output([img_input])
conv_output = conv_output[0]
grad_val = grad_val[0]
weights = np.mean(grad_val, axis=(0, 1))
grad_cam = np.zeros(dtype=np.float32, shape=conv_output.shape[0:2])
for k, w in enumerate(weights):
grad_cam += w * conv_output[:, :, k]
# RELU
grad_cam = np.maximum(grad_cam, 0)
grad_cam = cv2.resize(grad_cam, (224, 224))
# Guided grad-CAM
register_gradient()
guided_model, activation_layer = modify_backprop(resnet, 'GuidedBackProp',
args.checkpoint_path,
args.main_name)
saliency_fn = compile_saliency_function(guided_model, activation_layer)
saliency = saliency_fn([img_input, 0])
gradcam = saliency[0] * grad_cam[..., np.newaxis]
gradcam = deprocess_image(gradcam)
# grad_cam = ndimage.zoom(grad_cam, (32, 32), order=1)
plt.subplot(1, 2, 1)
plt.imshow(img, alpha=0.8)
plt.imshow(grad_cam, cmap='jet', alpha=0.5)
plt.axis('off')
plt.subplot(1, 2, 2)
plt.imshow(gradcam, cmap='jet', alpha=0.5)
plt.axis('off')
plt.show()
def sample2():
name2 = "here is index2"
pic = request.files['img_file']
#pic = urllib.request.urlopen(url).read()
#print("pic",pic)
# img_bin = io.BytesIO(pic)
# img =Image.open(img_bin)
test = img_to_array(
load_img(pic, target_size=(28, 28), color_mode="grayscale"))
# for test in tests:
# x.append(test)
#test = x.reshape(1,2352)
x = np.asarray(test)
# print("shape===", x.shape)
#x= np.squeeze(x)
#x=np.asarray(x)
x = x.astype('float32')
#x=x.reshape(2352,1)
x = x / 255.0
x = (np.expand_dims(x, 0))
#print("x;",x)
# print('x:',x)
# x_test = io.BytesIO(x_test)
# # Pillowで開き、画像を保存する
# x_test = Image.open(x_test).convert('L')
# img_rows, img_cols = 28, 28
# x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
# x_test = x_test.astype('float32')
# x_test /= 255
#x_test = 1- np.array(x_test)
#X_test = x_test.reshape(1,784)
# test = x_test
# test = (np.expand_dims(test,0))
data = {'images': x.tolist()}
API_URL = 'https://root67.herokuapp.com/'
res = requests.post(API_URL, json=data)
result = []
for v in res:
try:
js = json.loads(v)
result.append(js['data'])
except Exception:
pass
result = int(result[0]['prediction'][0])
#print("もしかしたら：",result,"？かも")
return render_template("index2.html", result=result, name2=name2)
def get_images_array_from_path_list(img_path_list, image_size=(224, 224)):
# img_array : (N2, N1)
# expand_dims : (N2, N1) => (1, N2, N1)
# concatenate : (1, N2, N1) + (1, N2, N1) + ... => (1, N3, N2, N1)
img_list = [
np.expand_dims(img_to_array(load_img(img_path,
target_size=image_size)),
axis=0) for img_path in img_path_list
]
return np.concatenate(img_list, axis=0)
def image_generate(image, name, directory):
#画像をkerasの形式にする
img = load_img(image)
x = img_to_array(img)
x = x.reshape((1, ) + x.shape)
g = datagen.flow(x, batch_size=1, save_to_dir=directory, save_format='jpg')
#画像を水増し
for i in range(5):
batch = g.next()
def load_imgs(self,img_paths, target_size=(224, 224)):
"""画像ファイルのパスのリストから、配列のバッチを返す"""
_load_img = lambda x: img_to_array(
load_img(x, target_size=target_size)
)
img_list = [
np.expand_dims(_load_img(img_path), axis=0)
for img_path in img_paths
]
return np.concatenate(img_list, axis=0)
def process(self, path_label_tuple):
"""
Load image from gcs
Yields: Image, categorical label
"""
path, label = path_label_tuple
with self.file_system.open(path) as img_file:
img = np.array(
load_img(img_file, target_size=(self.img_size, self.img_size)))
yield img, label
def load_img(path):
"""
this method is used to read image data from previous path for vgg19
:param path: image file
:return: np.array object with shape(1,224,224,3) , ordered by RGB->BGR ,centering zero
"""
img = image.load_img(path, target_size=(224, 224))
arr = np.expand_dims(image.img_to_array(img), axis=0)
arr = preprocess_input(arr)
return arr
def create_img_tensor(test_dir):
img_path = os.path.join(test_dir, r'cats\cat.{}.jpg'.format(random.randint(1501, 1999)))
plt.imshow(plt.imread(img_path))
img = image.load_img(img_path, target_size=(150, 150))
img_tensor = image.img_to_array(img)
img_tensor = np.expand_dims(img_tensor, axis=0)
img_tensor /= 255.
#plt.imshow(img_tensor[0])
return img_tensor
def readImages(folder_path):
images = []
for i in range(len(folder_path)):
img = load_img(folder_path[i], target_size=(224, 224, 3))
# img = Image.open(folder_path[i])
img = np.array(img, dtype=np.float32)
images.append(img)
images = np.array(images)
return images
def retrieve_similar_images(given_image, cosine_similarity_df):
print('original image:')
original = load_img(given_image, target_size=(imgs_width, imgs_height))
plt.imshow(original)
plt.show()
print('most similar products........')
closest_imgs = cosine_similarity_df[given_image].sort_values(
ascending=False)[1:nb_closest_images + 1].index
closest_imgs_scores = cosine_similarity_df[given_image].sort_values(
ascending=False)[1:nb_closest_images + 1]
for i in range(0, len(closest_imgs)):
original = load_img(closest_imgs[i],
target_size=(imgs_width, imgs_height))
plt.imshow(original)
plt.show()
print('similarity score:', closest_imgs_scores[i])
def predict(file):
x = load_img(file, target_size=(longitud, altura))
x = img_to_array(x)
x = np.expand_dims(x, axis=0)
print("Prediction..")
array = cnn.predict(x) ## [1,0,0]
result = array[0]
answer = np.argmax(result)
print(predicciones[answer])
return answer
def read_and_prep_images(train1_file,
img_height=image_size,
img_width=image_size):
imgs = [
load_img(img_path, target_size=(img_height, img_width))
for img_path in train1_file
]
img_array = np.array([img_to_array(img) for img in imgs])
output = preprocess_input(img_array)
return (output)
def read_and_prep_image(image_path, model_name, img_width=IMAGE_WIDTH, img_height=IMAGE_HEIGHT):
image_raw = load_img(image_path, target_size=(img_height, img_width))
image_array = np.array([img_to_array(image_raw)])
if model_name == "ResNet50":
image = preprocess_input_ResNet50(image_array)
elif model_name == "VGG16":
image = preprocess_input_VGG16(image_array)
elif model_name == "VGG19":
image = preprocess_input_VGG19(image_array)
return image
def run():
my_model = load_model(filepath=model_path)
testing_car_df = csv_to_dataFrame(testing_df_path=testing_df)
img_name_list = list()
# instantiate a dictionary and fill up keys
for row_index, row in testing_car_df.iterrows():
image_name = row['ImageFilename']
# img_dict[image_name] = list()
img_name_list.append(image_name)
column_list = ['ImageFilename', classifier]
result_df = pd.DataFrame(columns=column_list)
df_list = list()
for img_name in img_name_list:
test_img_path = join(img_folder_dir, img_name)
# img_path = 'glass_model_test_sample/no_damage_0.jpg'
img = image.load_img(path=test_img_path,
grayscale=False,
target_size=(224, 224))
x = image.img_to_array(img)
# print('image_to_array: ', x)
# print(x.shape)
x = np.expand_dims(x, axis=0)
# print('np_expand_dims', x)
# print(x.shape) # current shape -- (1, 224, 224, 3)
## the image is now prepared -- for CNN, input must be a 4-D tensor [batch_size, width, height, channel]
## channel -- 1: gray scale, 3: RGB(red, green glue)
prediction = my_model.predict(x)
print(test_img_path)
print('NoGlassDmg', ' ', 'YesGlassDmg')
print(prediction)
print(prediction[0][0], prediction[0][1])
no_val = prediction[0][0]
yes_val = prediction[0][1]
if no_val >= yes_val:
eval = 'no'
else:
eval = 'yes'
one_row_data = [img_name, eval]
one_row_df = pd.DataFrame(data=[one_row_data], columns=column_list)
result_df = result_df.append(one_row_df, ignore_index=True)
csv_output(df=result_df, csv_output_name='bumper_damage_result.csv')
def create_test_data(self):
sar_filename = self.load_sample(
sar_dir=
'/emwusr/xianzhengshi/2020_test/20200404/sar_seg/sar_256_test')
test_data = np.ndarray(
(len(sar_filename), self.img_cols, self.img_rows, 3))
for num, name in enumerate(sar_filename):
img = load_img(name)
img = img_to_array(img)
test_data[num] = img
return test_data
def preprocess_image(im_path, im_size, model_name):
im = image.load_img(im_path, target_size=(im_size[0], im_size[1]))
im = image.img_to_array(im)
im = np.expand_dims(im, axis=0)
if model_name == 'inception_v3':
im = inception_v3.preprocess_input(im)
elif model_name == 'resnet50':
im = resnet50.preprocess_input(im)
elif model_name == 'vgg16':
im = vgg16.preprocess_input(im)
return im
def densenet_extract_feat(self, img_path):
img = image.load_img(img_path,
target_size=(self.input_shape[0],
self.input_shape[1]))
img = image.img_to_array(img)
img = np.expand_dims(img, axis=0)
img = preprocess_input_densenet(img)
feat = self.model_densenet.predict(img)
# print(feat.shape)
norm_feat = feat[0] / LA.norm(feat[0])
return norm_feat
def _get_elephant(target_size):
# For models that don't include a Flatten step,
# the default is to accept variable-size inputs
# even when loading ImageNet weights (since it is possible).
# In this case, default to 299x299.
if target_size[0] is None:
target_size = (299, 299)
test_image = data_utils.get_file('elephant.jpg', TEST_IMAGE_PATH)
img = image.load_img(test_image, target_size=tuple(target_size))
x = image.img_to_array(img)
return np.expand_dims(x, axis=0)
def get_image(size):
"""Returns an image loaded into an np.ndarray with dims [1, size, size, 3].
Args:
size: Size of image.
Returns:
np.ndarray.
"""
img_filename = _resource_loader.get_path_to_datafile(
"testdata/grace_hopper.jpg")
img = image.load_img(img_filename, target_size=(size, size))
img_array = image.img_to_array(img)
img_array = np.expand_dims(img_array, axis=0)
return img_array
def test():
import os
import numpy as np
from PIL import Image
from tensorflow.python.keras.preprocessing.image import load_img
from models import Darknet19Encoder, Darknet19Decoder
inputShape = (256, 256, 3)
batchSize = 10
latentSize = 100
img = load_img(os.path.join('..','images', 'img.jpg'), target_size=inputShape[:-1])
img.show()
img = np.array(img, dtype=np.float32) / 255 - 0.5
img = np.array([img]*batchSize) # make fake batches to improve GPU utilization
# This is how you build the autoencoder
encoder = Darknet19Encoder(inputShape, batchSize, latentSize, 'bvae', beta=69, capacity=15, randomSample=True)
decoder = Darknet19Decoder(inputShape, batchSize, latentSize)
bvae = AutoEncoder(encoder, decoder)
bvae.ae.compile(optimizer='adam', loss='mean_absolute_error')
while True:
bvae.ae.fit(img, img,
epochs=100,
batch_size=batchSize)
# example retrieving the latent vector
latentVec = bvae.encoder.predict(img)[0]
print(latentVec)
pred = bvae.ae.predict(img) # get the reconstructed image
pred[pred > 0.5] = 0.5 # clean it up a bit
pred[pred < -0.5] = -0.5
pred = np.uint8((pred + 0.5)* 255) # convert to regular image values
pred = Image.fromarray(pred[0])
pred.show() # display popup
def load_imgs(img_paths, target_size):
list_imgs = [img_to_array(load_img(path, target_size=target_size))
for path in img_paths]
return np.array(list_imgs)
def predictMultipleImages(self, sent_images_array, result_count_per_image=2, input_type="file"):
"""
'predictMultipleImages()' function is used to predict more than one image by receiving the following arguments:
* input_type , the type of inputs contained in the parsed array. Acceptable values are "file", "array" and "stream"
* sent_images_array , an array of image file paths, image numpy array or image file stream
* result_count_per_image (optionally) , the number of predictions to be sent per image, which must be whole numbers between
1 and 1000. The default is 2.
This function returns an array of dictionaries, with each dictionary containing 2 arrays namely 'prediction_results' and 'prediction_probabilities'. The 'prediction_results'
contains possible objects classes arranged in descending of their percentage probabilities. The 'prediction_probabilities'
contains the percentage probability of each object class. The position of each object class in the 'prediction_results'
array corresponds with the positions of the percentage possibilities in the 'prediction_probabilities' array.
:param input_type:
:param sent_images_array:
:param result_count_per_image:
:return output_array:
"""
output_array = []
for image_input in sent_images_array:
prediction_results = []
prediction_probabilities = []
if (self.__modelLoaded == False):
raise ValueError("You must call the loadModel() function before making predictions.")
else:
if (self.__modelType == "squeezenet"):
from .imagenet_utils import preprocess_input, decode_predictions
if (input_type == "file"):
try:
image_to_predict = image.load_img(image_input, target_size=(self.__input_image_size, self.__input_image_size))
image_to_predict = image.img_to_array(image_to_predict, data_format="channels_last")
image_to_predict = np.expand_dims(image_to_predict, axis=0)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have set a path to an invalid image file.")
elif (input_type == "array"):
try:
image_input = Image.fromarray(np.uint8(image_input))
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong numpy array for the image")
elif (input_type == "stream"):
try:
image_input = Image.open(image_input)
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong stream for the image")
model = self.__model_collection[0]
prediction = model.predict(image_to_predict, steps=1)
try:
predictiondata = decode_predictions(prediction, top=int(result_count_per_image))
for results in predictiondata:
countdown = 0
for result in results:
countdown += 1
prediction_results.append(str(result[1]))
prediction_probabilities.append(str(result[2] * 100))
except:
raise ValueError("An error occured! Try again.")
each_image_details = {}
each_image_details["predictions"] = prediction_results
each_image_details["percentage_probabilities"] = prediction_probabilities
output_array.append(each_image_details)
elif (self.__modelType == "resnet"):
model = self.__model_collection[0]
from .imagenet_utils import preprocess_input, decode_predictions
if (input_type == "file"):
try:
image_to_predict = image.load_img(image_input, target_size=(self.__input_image_size, self.__input_image_size))
image_to_predict = image.img_to_array(image_to_predict, data_format="channels_last")
image_to_predict = np.expand_dims(image_to_predict, axis=0)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have set a path to an invalid image file.")
elif (input_type == "array"):
try:
image_input = Image.fromarray(np.uint8(image_input))
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong numpy array for the image")
elif (input_type == "stream"):
try:
image_input = Image.open(image_input)
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong stream for the image")
prediction = model.predict(x=image_to_predict, steps=1)
try:
predictiondata = decode_predictions(prediction, top=int(result_count_per_image))
for results in predictiondata:
countdown = 0
for result in results:
countdown += 1
prediction_results.append(str(result[1]))
prediction_probabilities.append(str(result[2] * 100))
except:
raise ValueError("An error occured! Try again.")
each_image_details = {}
each_image_details["predictions"] = prediction_results
each_image_details["percentage_probabilities"] = prediction_probabilities
output_array.append(each_image_details)
elif (self.__modelType == "densenet"):
model = self.__model_collection[0]
from .DenseNet.densenet import preprocess_input, decode_predictions
from .DenseNet.densenet import DenseNetImageNet121
if (input_type == "file"):
try:
image_to_predict = image.load_img(image_input, target_size=(self.__input_image_size, self.__input_image_size))
image_to_predict = image.img_to_array(image_to_predict, data_format="channels_last")
image_to_predict = np.expand_dims(image_to_predict, axis=0)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have set a path to an invalid image file.")
elif (input_type == "array"):
try:
image_input = Image.fromarray(np.uint8(image_input))
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong numpy array for the image")
elif (input_type == "stream"):
try:
image_input = Image.open(image_input)
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong stream for the image")
prediction = model.predict(x=image_to_predict, steps=1)
try:
predictiondata = decode_predictions(prediction, top=int(result_count_per_image))
for results in predictiondata:
countdown = 0
for result in results:
countdown += 1
prediction_results.append(str(result[1]))
prediction_probabilities.append(str(result[2] * 100))
except:
raise ValueError("An error occured! Try again.")
each_image_details = {}
each_image_details["predictions"] = prediction_results
each_image_details["percentage_probabilities"] = prediction_probabilities
output_array.append(each_image_details)
elif (self.__modelType == "inceptionv3"):
model = self.__model_collection[0]
from imageai.Prediction.InceptionV3.inceptionv3 import InceptionV3, preprocess_input, \
decode_predictions
if (input_type == "file"):
try:
image_to_predict = image.load_img(image_input, target_size=(self.__input_image_size, self.__input_image_size))
image_to_predict = image.img_to_array(image_to_predict, data_format="channels_last")
image_to_predict = np.expand_dims(image_to_predict, axis=0)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have set a path to an invalid image file.")
elif (input_type == "array"):
try:
image_input = Image.fromarray(np.uint8(image_input))
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong numpy array for the image")
elif (input_type == "stream"):
try:
image_input = Image.open(image_input)
image_input = image_input.resize((self.__input_image_size, self.__input_image_size))
image_input = np.expand_dims(image_input, axis=0)
image_to_predict = image_input.copy()
image_to_predict = np.asarray(image_to_predict, dtype=np.float64)
image_to_predict = preprocess_input(image_to_predict)
except:
raise ValueError("You have parsed in a wrong stream for the image")
prediction = model.predict(x=image_to_predict, steps=1)
try:
predictiondata = decode_predictions(prediction, top=int(result_count_per_image))
for results in predictiondata:
countdown = 0
for result in results:
countdown += 1
prediction_results.append(str(result[1]))
prediction_probabilities.append(str(result[2] * 100))
except:
raise ValueError("An error occured! Try again.")
each_image_details = {}
each_image_details["predictions"] = prediction_results
each_image_details["percentage_probabilities"] = prediction_probabilities
output_array.append(each_image_details)
return output_array