def infer(self, image_file, autocrop=False):
"""Run inference on the given image"""
# Load and preprocess image
image = cv2.imread(image_file, cv2.IMREAD_GRAYSCALE)
if autocrop:
image, _ = auto_body_crop(image)
image = cv2.resize(image, (self.input_width, self.input_height), cv2.INTER_CUBIC)
image = image.astype(np.float32) / 255.0
image = np.expand_dims(np.stack((image, image, image), axis=-1), axis=0)
# Create feed dict
feed_dict = {IMAGE_INPUT_TENSOR: image, TRAINING_PH_TENSOR: False}
# Run inference
graph, sess, saver = self.load_graph()
with graph.as_default():
# Load checkpoint
self.load_ckpt(sess, saver)
# Run image through model
class_, probs = sess.run([CLASS_PRED_TENSOR, CLASS_PROB_TENSOR], feed_dict=feed_dict)
print('\nPredicted Class: ' + CLASS_NAMES[class_[0]])
print('Confidences:' + ', '.join(
'{}: {}'.format(name, conf) for name, conf in zip(CLASS_NAMES, probs[0])))
print('**DISCLAIMER**')
print('Do not use this prediction for self-diagnosis. '
'You should check with your local authorities for '
'the latest advice on seeking medical assistance.')
def infer(self, image_file, autocrop=False):
"""Run inference on the given image"""
# Load and preprocess image
image = cv2.imread(image_file, cv2.IMREAD_GRAYSCALE)
if autocrop:
image, _ = auto_body_crop(image)
image = cv2.resize(image,
(self.input_width, self.input_height),
cv2.INTER_CUBIC)
image = image.astype(np.float32) / 255.0
image = np.expand_dims(np.stack((image, image, image),
axis=-1),
axis=0)
# Create feed dict
feed_dict = {
IMAGE_INPUT_TENSOR: image,
TRAINING_PH_TENSOR: False
}
# Run inference
graph, sess, saver = self.load_graph()
with graph.as_default():
# Load checkpoint
self.load_ckpt(sess, saver)
# Run image through model
class_, probs = sess.run(
[CLASS_PRED_TENSOR, CLASS_PROB_TENSOR],
feed_dict=feed_dict)
print('\nPredicted Class: ' + CLASS_NAMES[class_[0]])
print('Confidences:' + ', '.join(
'{}: {}'.format(name, conf)
for name, conf in zip(CLASS_NAMES, probs[0])))
N = probs[0][0]
P = probs[0][1]
C = probs[0][2]
result = CLASS_NAMES[class_[0]]
print('**DISCLAIMER**')
print(
'Do not use this prediction for self-diagnosis. '
'You should check with your local authorities for '
'the latest advice on seeking medical assistance.')
confidence = {
'normal': str("%.2f" % (N * 100)),
'pneumonia': str("%.2f" % (P * 100)),
'covid': str("%.2f" % (C * 100))
}
detection = {
'detections': {
'prediction': result,
'confidence': confidence,
'imageType': 'CT Scan'
}
}
my_logger.info(
'CT Scan Detection : {}'.format(detection))
try:
if (not len(firebase_admin._apps)):
cred = credentials.Certificate(
credential_json_file)
fa = firebase_admin.initialize_app(
cred, {
"databaseURL": databaseURL,
'storageBucket': storageBucket
})
fc = firebase_admin.firestore.client(fa)
db = firestore.client
doc_ref = db.collection(
u'stripe_customers/{0}/results'.format(
userId)).document(currentTime)
doc_ref.update(detection)
my_logger.error(
'CT Scan Detection : Saved to firestore and message acknowledged'
)
else:
print('alredy initialize')
db = firestore.client()
doc_ref = db.collection(
u'stripe_customers/{0}/results'.format(
userId)).document(currentTime)
doc_ref.update(detection)
my_logger.error(
'CT Scan Detection : Saved to firestore and message acknowledged'
)
except Exception as e:
print(e)
my_logger.error('CT Scan Detection : NOT Saved', e)
def uploaded_ct():
if request.method == 'POST':
# check if the post request has the file part
if 'file' not in request.files:
flash('No file part')
return redirect('upload.html')
file = request.files['file']
# if user does not select file, browser also
# submit a empty part without filename
if file.filename == '':
flash('No file selected for uploading')
return redirect('upload.html')
if file:
filename = secure_filename(file.filename)
file_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
file.save(file_path)
# Load and preprocess image
origin_im = cv2.imread(file_path) # read file
origin_im = cv2.cvtColor(origin_im, cv2.COLOR_BGR2RGB)
print(origin_im.shape)
h, w, c = origin_im.shape
image = cv2.imread(file_path, cv2.IMREAD_GRAYSCALE) # read file
image, _ = auto_body_crop(image)
image = cv2.resize(image, (512, 512), cv2.INTER_CUBIC)
image = image.astype(np.float32) / 255.0
image = np.expand_dims(np.stack((image, image, image), axis=-1),
axis=0)
with tf.Graph().as_default() as graph:
tf.import_graph_def(
graph_def1,
input_map=None,
return_elements=None,
name="",
#op_dict=None,
#producer_op_list=None
)
image_tensor = graph.get_tensor_by_name("Placeholder:0")
pred_tensor = graph.get_tensor_by_name("softmax_tensor:0")
TRAINING_PH_TENSOR = graph.get_tensor_by_name("is_training:0")
sess = tf.Session(graph=graph)
gradCam = GradCAM(graph=graph,
classes=[0, 1, 2],
outLayer="softmax_tensor:0",
targetLayer="resnet_model/add_15:0")
grads = gradCam.compute_grads()
size_upsample = (w, h)
# Create feed dict
feed_dict = {image_tensor: image, TRAINING_PH_TENSOR: False}
pred = sess.run(pred_tensor, feed_dict)
output, grads_val = sess.run(
[gradCam.target, grads[pred.argmax(axis=1)[0]]], feed_dict)
cam3 = generate_cam(output[0], grads_val[0], size_upsample)
# Overlay cam on image
cam3 = np.uint8(255 * cam3)
cam3 = cv2.applyColorMap(cam3, cv2.COLORMAP_JET)
new_im = cam3 * 0.3 + origin_im * 0.5
filename1 = my_random_string(10) + filename
filename2 = my_random_string(12) + filename
os.rename(file_path,
os.path.join(app.config['UPLOAD_FOLDER'], filename1))
cv2.imwrite(
os.path.join(app.config['UPLOAD_FOLDER'], filename2),
new_im)
print("GradCAM CT image saved ")
print(pred)
inv_mapping = {0: 'normal', 1: 'pneumonia', 2: 'COVID-19'}
pred_class = inv_mapping[pred.argmax(axis=1)[0]]
pred_proba = "{:.2f}".format((pred.max(axis=1)[0]) * 100)
print(pred_proba)
print(pred_class)
result = pred_class.capitalize()
return render_template('results_ct.html',
result=result,
probability=pred_proba,
imagesource='/assets/images/' + filename1,
imagesource1='/assets/images/' + filename2)