def post(self):
args = parser.parse_args()
image = np.array(json.loads(args['image']))
image = process_image(image)
with graph.as_default():
prediction = model.predict(image)
return prediction.tolist()[0]
def init_input():
image = process_new(model_params, False)
with graph.as_default():
preds = model.predict(image)
preds = preds[0]
preds = [round(val, 2) for val in preds[:3]]
image_loc = "../static/images/demo2.jpg"
return render_template("pages/input.html", image=image_loc, preds=preds)
def classify_image(image):
"""Call ResXception to return emotion probabilities
from Keras model.
@param image: RGB cropped face image
@return: map from emotion to probability
"""
nn_input_shape = nn.input_shape[1:3]
gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
gray_image = cv2.resize(gray_image, (nn_input_shape))
gray_image = clean_image(gray_image)
gray_image = np.expand_dims(gray_image, 0)
gray_image = np.expand_dims(gray_image, -1)
with graph.as_default():
emotion_proba = nn.predict(gray_image)
emotion_map = index_to_emotion(emotion_proba[0].tolist())
return emotion_map
def upload_image():
if request.method == 'POST':
if request.files:
image = request.files['imgInp']
if allow_image(image.filename, ["JPEG", "JPG", "PNG"]):
# Image preprocessing and classification.
global sess
global graph
global new_model
test_img = process_img(image, 150, 32)
with graph.as_default():
set_session(sess)
prediction = new_model.predict(test_img, verbose=1)[0][0]
prob = 100 * round(prediction, 3)
classes = ['Normal', 'Pneumonia']
pred_class = classes[int(round(prediction))]
c = [100 - prob, prob]
imagem = get_encoded_image(c)
# Memory release
gc.collect()
# The result.html page will be rendered with 3 objects:
# - The class of uploaded image;
# - Probability of 'pneumonia' class;
# - Generated donut plot containing the class probabilities within the HTML code.
return render_template(
'result.html',
classification=pred_class,
probability=prob,
img_code=Markup(
'<img src=data:image/png;base64,{} style="height:250px; width:auto;" alt="Responsive image">'
.format(imagem)))
else:
# In case file is no choosen or file format is invalid.
feedback = "File format not allowed."
return render_template("diagnosis.html", feedback=feedback)
return render_template('diagnosis.html')
def upload_file():
if request.method == 'POST':
f = request.files['file']
path = os.path.join(app.config['UPLOAD_FOLDER'], f.filename)
f.save(path)
img = image.load_img(path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = mobilenet.preprocess_input(x)
global graph
with graph.as_default():
prediction = model.predict(x)
results = imagenet_utils.decode_predictions(prediction)
rst = results[0][0][1]
return render_template('uploaded.html',
title='Success',
predictions=rst,
user_image=f.filename)
def download():
url = request.args['url']
filename = request.args.get('filename', 'image.png')
r = requests.get(url)
imgnamedate = datetime.datetime.now().strftime("%Y%m%d%H%M%S")
with app.open_instance_resource(imgnamedate + '.jpg', 'wb') as f:
f.write(r.content)
imagepath = app.open_instance_resource(imgnamedate + '.jpg')
img = image.load_img(imagepath, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = mobilenet.preprocess_input(x)
global graph
with graph.as_default():
prediction = model.predict(x)
prediction = model.predict(x)
results = imagenet_utils.decode_predictions(prediction)
userimage = imgnamedate + '.jpg'
rst = results[0][0][1]
return render_template('uploaded.html',
title='Success',
predictions=rst,
user_image=userimage)