def train():
print("Method", "facerec_service.train")
align.align_dataset_mtcnn.main(
align.align_dataset_mtcnn.parse_arguments(
[IMAGE_ORIGINAL_DIR, IMAGE_PREPARED_DIR]))
classifier.main(
classifier.parse_arguments(
['TRAIN', IMAGE_PREPARED_DIR, MODEL_PATH, CLASSIFIER_PATH]))
def predict():
animals = ['Assult', 'Normal', 'Arrest', 'Explosion', 'Vandalism']
file = file_upload("Select a video:", accept="video/*")
start = time.time()
with put_loading():
f = open('temp.mp4', 'wb')
f.write(file['content'])
result, list_final = main('temp.mp4')
# tx = "Prediction: "+str(result)
#Prediction: "+str(result)
img = open('output_img.jpg', 'rb').read()
style(put_text('Middle frame from Video'), 'text-align: center')
style(put_image(img, width='500px'),
'display: block; margin-left: auto; margin-right: auto')
fig = go.Figure([go.Bar(x=animals, y=list_final)])
html = fig.to_html(include_plotlyjs="require", full_html=False)
put_text('\n')
style(put_text('Prediction Graph'), 'text-align: center')
style(put_html(html), 'margin: auto')
end = time.time()
print("Time elapsed:", end - start)
print(list_final)
# put_text(tesxt)
style(put_text('Prediction:'), 'text-align: center')
style(put_text(str(result)), 'font-size: 200%;text-align: center')
def browseFiles():
filename = filedialog.askopenfilename(initialdir = "/home/mittooji/Downloads/",
title = "Select a File",
filetypes = (("Video files","*.mp4*"),
("all files","*.*")))
prediction = main(filename)
label_file_explorer.configure(text="Anomaly Label: "+prediction)
import os
import classifier
if __name__ == '__main__':
# Project root path to make it easier to reference files
project_root_path = os.path.join(os.path.abspath(__file__), "..\\..")
# Feel free to replace this and use actual commandline args instead, the main method will still work
args = lambda: None
args.data_dir = project_root_path + '\\training_data_aligned'
args.seed = None
args.use_split_dataset = False
args.model = project_root_path + '\\facenet_model\\20170512-110547\\20170512-110547.pb'
args.mode = 'TRAIN'
args.batch_size = 460
args.image_size = 160
args.classifier_filename = project_root_path + '\\trained_classifier\\newglint_classifier.pkl'
classifier.main(args)
def train():
sys.argv[1:] = [var3.get(), var1.get(), var2.get()]
classifier.main(classifier.parse_arguments(sys.argv[1:]))
from tkinter.messagebox import showinfo
tk.messagebox.showinfo(title="tips", message="训练完毕")
def main():
return classifier.main((parameter))
#!/usr/bin/env python
import classifier
import sys
if __name__ == "__main__":
classifier.main(args=sys.argv[1:])
classify = cv2.CascadeClassifier(
"C:\\Users\\ramakriy\\AppData\\Local\\Programs\\Python\\Python35\\Lib\\site-packages\\cv2\\data\\haarcascade_frontalface_alt.xml"
)
webcam = cv2.VideoCapture(0)
while True:
(rval, im) = webcam.read()
im = cv2.flip(im, 1, 0)
mini = cv2.resize(im, (int(im.shape[1] / size), int(im.shape[0] / size)))
faces = classify.detectMultiScale(mini)
# Draw rectangles around each face
for f in faces:
(x, y, w, h) = [v * size for v in f]
cv2.rectangle(im, (x, y), (x + w, y + h), (0, 255, 0), 4)
sub_face = im[y:y + h, x:x + w]
FaceFileName = "test.jpg" #Saving the current image from the webcam for testing.
cv2.imwrite(FaceFileName, sub_face)
text = classifier.main(
FaceFileName) # Getting the Result from the Classification Result.
text = text.title()
font = cv2.FONT_HERSHEY_TRIPLEX
cv2.putText(im, text, (x + w, y), font, 1, (0, 0, 255), 2)
# Show the image
cv2.imshow('Capture', im)
key = cv2.waitKey(10)
# if Esc key is press then break out of the loop
if key == 27:
break
def test_one_instance(self, label_type, x_tuple, percent_tuple, run_num):
x, x_value = x_tuple
p, percent_sim_data = percent_tuple
budget = x_value
self.target_env.env.set_to_training_set()
# Get data from oracle feedback
num_seen_states, reviewed_blindspots = review.main(self.filenames, self.save_dir, self.target_env, label_type, budget, percent_sim_data, self.max_states, 0, self.percentile)
if self.max_states == -1 or self.max_states < num_seen_states:
self.max_states = num_seen_states
prediction_probs_skene = None
prediction_classes_skene = None
for i in self.classifier_baselines:
predicted_filename = os.path.join(self.save_dir, i+".csv")
# Run approach given oracle data
accuracy, mean_squared_error, error1s = self.run_approach(i, self.filenames["data"], predicted_filename, reviewed_blindspots, label_type)
if i in self.estimation_baselines:
self.estimation_results[label_type[0]]["accuracy"][i][x,run_num] = accuracy
self.estimation_results[label_type[0]]["error"][i][x,run_num] = mean_squared_error
self.estimation_results[label_type[0]]["error1s"][i][x,run_num] = error1s
results, self.data_sizes[label_type[0]], prediction_probs, prediction_classes = classifier.main(self.save_dir, predicted_filename, self.filenames["true_sim"], label_type[0], i, self.classifier_metrics, self.filenames["sim_on_real"])
if i == "dawid_skene":
prediction_probs_skene = prediction_probs
prediction_classes_skene = prediction_classes
for metric in self.classifier_metrics:
for t in self.test_data_list:
self.classifier_results[label_type[0]][metric][t][i][x,run_num] = results[t][metric]
self.target_env.env.set_to_testing_set()
for i in self.oracle_in_loop_baselines:
probs = None
classes = None
if i == "model_query":
probs = prediction_probs_skene
classes = prediction_classes_skene
# Run oracle-in-the-loop evaluation
avg_reward, percent_queries = self.oracle_in_loop_eval(self.target_env, i, probs, classes)
self.oracle_in_loop_results[label_type[0]]["avg_reward"][i][x,run_num] = avg_reward
self.oracle_in_loop_results[label_type[0]]["percent_queries"][i][x,run_num] = percent_queries