def sign_up():
""" Save new user info into Mongo Database """
form = SignUp(csrf_enabled=False)
name = []
phone = []
email = []
occupation = []
age = []
description = []
unique_id = []
if request.method == "POST":
name.append(form.name.data)
phone.append(form.phone.data)
email.append(form.email.data)
occupation.append(form.occupation.data)
age.append(form.age.data)
description.append(form.description.data)
unique_id.append(form.unique_id.data)
# Save new user's image for retraining the face recognition model
save_image_for_training(unique_id[0], name[0])
new_entry = pd.DataFrame()
new_entry["Name"] = name
new_entry["Phone"] = phone
new_entry["Email"] = email
new_entry["Occupation"] = occupation
new_entry["Age"] = age
new_entry["Description"] = description
new_entry["Unique ID"] = unique_id
# Create new collection for users in app database and insert data
collection = "users"
db_cm = db[collection]
data = new_entry.to_dict(orient='records')
db_cm.insert(data)
try:
# Retrain model with new users
print("Training KNN classifier...")
train("static/train_test/train",
model_save_path="static/models/trained_knn_model.clf",
n_neighbors=None)
print("Training complete!")
except:
print("Oh no, unable to train model.")
status = "Agent assignment complete."
return render_template("index.html", form=form, status=status)
else:
return render_template("index.html", form=form)
def acquaintence_identification(uploaded_file):
try:
# Retrain model with new users
print("Training KNN classifier...")
train("static/train_test/train", model_save_path="static/models/trained_knn_model.clf", n_neighbors=None)
print("Training complete!")
except:
return render_template("search.html", status="Oops, no one is in here. Be the first to sign up! (:")
print("Looking for faces in {}".format(uploaded_file))
# Find all people in the image using a trained classifer model_path
# Note: You can pass in either a classifier file name or a classifier model isntance
predictions = predict(uploaded_file, model_path="static/models/trained_knn_model.clf")
# Print results on the console
for name, (top, right, bottom, left) in predictions:
print("-Found {} at ({}, {})".format(name, left, top))
if name == "unknown":
return render_template("search.html", status="Sorry, unable to identify them :( Please try another photo.")
else:
collection = db.users
query = collection.find({"Unique ID": name})
for q in query:
user_dict = {}
user_dict['Name'] = q['Name']
user_dict['Phone'] = q['Phone']
user_dict['Email'] = q['Email']
user_dict['Occupation'] = q['Occupation']
user_dict['Description'] = q['Description']
print(user_dict)
# Display results overland on an image
# show_prediction_labels_on_image(uploaded_file, predictions)
print("Rendering template...")
return render_template("search.html", status="No one here.", user=user_dict, src_path=uploaded_file )
## Run and Check the correctness
path_to_training_images = 'data/training/images'
training_csv_file = 'data/training/steering_angles.csv'
path_to_testing_images = 'data/training/images'
testing_csv_file = 'data/training/steering_angles.csv'
time_limit = 600
#Setup timout handler - I think this will only work on unix based systems:
signal.signal(signal.SIGALRM, timeout_handler)
signal.alarm(time_limit)
try:
print("Training your network, start time: %s" % time.strftime("%H:%M:%S"))
NN = tp.train(path_to_images=path_to_training_images,
csv_file=training_csv_file)
except ValueError as ex:
pass
finally:
signal.alarm(0)
print("Ending Time: %s" % time.strftime("%H:%M:%S"))
print('Training Complete! \n')
print('Measuring performance...')
## Measure Performance:
data = np.genfromtxt(testing_csv_file, delimiter=',')
frame_nums = data[:, 0]
steering_angles = data[:, 1]
# train predict loop for each feat set
#for fset, fset_results in all_tgt_results[tgt_name][est_type].items():
for fset_name, fset_list in fset_dict.items():
zeit = time.time()
print(
'%s/%s/%s/%d: scoring used:' %
(tgt_name, est_type, fset_name, fsets_count_int), scoring)
#feat_train = fset_results.data['fset_list'] + gbl.clin_demog_feats
feat_train = fset_list + gbl.clin_demog_feats
ests, train_scores = train(
est_type=est_type,
task=subs.tgt_task,
X=subs.pat_frame_train_norm.loc[:, feat_train],
y=subs.pat_frame_train_y,
cv_folds=subs.cv_folds,
params=params['params'],
scoring=scoring,
thresh=thresh)
print(
'%s/%s/%s/%d: trained cv over grid:' %
(tgt_name, est_type, fset_name, fsets_count_int), train_scores)
pred_scores, pred_frames = pred(
est_type=est_type,
task=subs.tgt_task,
ests=ests,
X=subs.pat_frame_test_norm.loc[:, feat_train],
y=subs.pat_frame_test_y.iloc[:, 0],
scoring=scoring,