import argument_parser
import trainer
from auxiliary.my_utils import *
from functools import reduce
from auxiliary.model_atlasnet import *
import miou_shape
import useful_losses as loss
import my_utils
from my_utils import Max_k, Min_k
import ICP
import tqdm
from save_mesh_from_points_and_labels import *
import figure_2_3
import pdb
opt = argument_parser.parser()
my_utils.plant_seeds(randomized_seed=opt.randomize)
min_k = Min_k(opt.k_max_eval)
max_k = Max_k(opt.k_max_eval)
if opt.num_figure_3_4 > 0:
min_k_fig = Min_k(opt.num_figure_3_4)
max_k_fig = Max_k(opt.num_figure_3_4)
trainer = trainer.Trainer(opt)
trainer.build_dataset_train_for_matching()
trainer.build_dataset_test_for_matching()
trainer.build_network()
trainer.network.eval()
# =============DEFINE Criterions to evaluate======================================== #
def draw_face_direction(img, image_points, nose_end_point2D, direction):
cv2.putText(img, 'Yaw' + str(direction["Yaw"]), (20, 10), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
cv2.putText(img, 'Pitch' + str(direction["Pitch"]), (20, 25), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
cv2.putText(img, 'Roll' + str(direction["Roll"]), (20, 40), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
p1 = (int(image_points[0][0]), int(image_points[0][1]))
p2 = (int(nose_end_point2D[0][0][0]), int(nose_end_point2D[0][0][1]))
cv2.line(img, p1, p2, (255, 0, 0), 2)
return img
def send_udp(client, server_address, direction):
message = json.dumps(direction)
client.sendto(message.encode(), server_address)
def low_pass_filter(now, prev, k):
return np.array([do_low_pass_filter(x, y, k) for x, y in zip(now, prev)])
def do_low_pass_filter(now, prev, k):
return (1 - k) * prev + k * now
if __name__ == "__main__":
args = parser()
print(args)
main(args)