def main(model_path, input_source, SDCNet=False):
if os.path.isfile(input_source):
video_name = input_source.split("/")[-1]
video_path = input_source.replace(video_name,"")
print("Processing video {} in {}".format(colored(video_name, 'red'), colored(video_path, 'green')))
cap = cv2.VideoCapture(input_source)
else:
cap = VideoStream(input_source).start()
# --1.2 use initial setting to generate
# set label_indice
label_indice = np.arange(0.5, 22+0.5/2, 0.5)
add = np.array([1e-6, 0.05, 0.10, 0.15, 0.20,
0.25, 0.30, 0.35, 0.40, 0.45])
label_indice = np.concatenate((add, label_indice))
# init networks
label_indice = torch.Tensor(label_indice)
class_num = len(label_indice)+1
div_times = 2
if SDCNet:
net = SDCNet_VGG16_classify(
class_num, label_indice, load_weights=True)
else:
net = SSDCNet_classify(class_num, label_indice, div_times=div_times,
frontend_name='VGG16', block_num=5,
IF_pre_bn=False, IF_freeze_bn=False, load_weights=True,
parse_method='maxp')
if os.path.exists(model_path):
print("Adding Weights ....")
all_state_dict = torch.load(model_path,map_location=torch.device('cpu'))
net.load_state_dict(all_state_dict['net_state_dict'])
net.eval()
else:
print("Can't find trained weights!!")
exit()
first_frame = True
while True:
flag, image = cap.read()
output_image = np.copy(image)
if first_frame:
rois = cv2.selectROIs("frame", image, False, False)
first_frame = False
# print(roi)
if flag:
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
else:
break
sum = 0
for roi in rois:
roi_image = image[int(roi[1]):int(roi[1]+roi[3]),
int(roi[0]):int(roi[0]+roi[2])]
roi_image = cv2.resize(roi_image, (256, 256))
roi_image = np.transpose(roi_image, (2, 0, 1))
roi_image = torch.Tensor(roi_image[None, :, :, :])
# w = image.shape[-1]
# h = image.shape[-2]
# pad_w = 64 - w%64
# padding_left = int(pad_w/2)
# padding_right = pad_w - padding_left
# pad_h = 64 - h%64
# padding_top = int(pad_h/2)
# padding_bottom = pad_h - padding_top
# image = torch.nn.functional.pad(image, (padding_left, padding_right, padding_top, padding_bottom))
with torch.no_grad():
features = net(roi_image)
div_res = net.resample(features)
merge_res = net.parse_merge(div_res)
if SDCNet:
outputs = merge_res['div'+str(net.args['div_times'])]
else:
outputs = merge_res['div'+str(net.div_times)]
del merge_res
cv2.rectangle(output_image, (int(roi[0]), int(roi[1])), (int(
roi[0]+roi[2]), int(roi[1]+roi[3])), (255, 0, 0), thickness=3)
sum += int(outputs.sum().item())
cv2.putText(output_image, "{}".format(sum),
(30, 50), cv2.FONT_HERSHEY_PLAIN, 2,
(255, 0, 0), 3)
cv2.imshow("frame", output_image)
if cv2.waitKey(1) == ord('q'):
cap.release()
exit()
cap.release()
class Game:
def __init__(self):
# initialize game window, etc
pg.init()
pg.mixer.init()
self.screen = pg.display.set_mode((WIDTH, HEIGHT)) # 창 크기
pg.display.set_caption(TITLE)
pg.display.set_icon(pg.image.load('images/icon.png')) #set icon
self.blink_event = pg.event.Event(pg.USEREVENT)
self.load_data()
self.clock = pg.time.Clock()
self.playing = False
self.start = True
self.practice = True
def new(self):
# start a new game
self.clear = False
self.ending = False
self.hit_count = 0
self.blink_sign = list()
self.blink_data = list() # data list
self.blink_index = 0
self.blink_dataLen = 0 # data len
self.blink_action = 0
self.game_tick = 0
self.vs = VideoStream(self, device=0 + cv2.CAP_DSHOW,
model=self.CNN).start()
self.all_sprites = pg.sprite.Group()
self.blinks = pg.sprite.Group() # 깜빡임 지시자 sprite 그룹 생성
pg.mixer.music.load(os.path.join(self.snd_dir, METRONOME)) #배경음 로드
self.run()
def run(self):
self.ready_sound.play()
self.start_tick = pg.time.get_ticks()
self.load_blinkData()
#game loop
pg.mixer.music.play(loops=-1) #배경음 플레이 (loops 값 false = 반복, true = 한번)
self.playing = True
while self.playing:
self.clock.tick(FPS)
self.update()
self.draw()
self.events()
pg.mixer.music.fadeout(500) #배경음이 갑자기 꺼지지 않고 점점 꺼지게 함
def update(self):
#game loop - update
self.all_sprites.update()
self.game_tick = pg.time.get_ticks() - self.start_tick
# 끝
if self.ending == True:
self.playing = False
if self.hit_count == self.blink_dataLen - 1: #잠금 해제
self.clear_text()
sleep(0.5)
self.clear = True
def clear_text(self):
for i in range(8):
self.draw_text('!!패턴 일치!!', 60, BLUE, WIDTH / 2, HEIGHT / 2 - 100)
pg.display.update()
sleep(0.1)
self.draw_text('!!패턴 일치!!', 60, GRAY, WIDTH / 2, HEIGHT / 2 - 100)
pg.display.update()
sleep(0.1)
def events(self):
#game loop - events
for event in pg.event.get():
if event.type == pg.USEREVENT:
if self.blink_action < len(self.blink_sign):
position_of_bar = self.blink_sign[self.blink_action].rect.x
print(position_of_bar)
if position_of_bar >= (
WIDTH / 2 - 20) and position_of_bar <= (WIDTH / 2 +
20):
self.hit_count += 1
print('hit!')
self.blink_sign[self.blink_action].correct = 1
self.blink_action += 1
if self.hit_count == self.blink_dataLen - 1:
self.ending = True
elif event.type == pg.QUIT:
if self.playing:
self.playing = False
self.start = False
self.start = False
# 데이터를 불러오는 함수
def load_data(self):
self.dir = os.path.dirname(__file__)
#model
silence_tensorflow()
self.CNN = load_model('models/' + MODEL_VER)
#image
self.img_dir = os.path.join(self.dir, 'images')
self.bg_img = pg.image.load(os.path.join(self.img_dir, 'main.png'))
self.team_img1 = pg.image.load(os.path.join(self.img_dir, 'team1.jpg'))
self.team_img2 = pg.image.load(os.path.join(self.img_dir, 'team2.jpg'))
self.end_img = pg.image.load(os.path.join(self.img_dir, 'ending.png'))
self.menu_select = pg.image.load(
os.path.join(self.img_dir, 'menu_select.png'))
self.font_name = pg.font.match_font(FONT_NAME) #FONT_NMAE과 맞는 폰트를 검색
self.fnt_dir = os.path.join(self.dir, 'font')
self.gg_font = os.path.join(self.fnt_dir, GG)
#sound(효과음)
self.snd_dir = os.path.join(self.dir, 'sound')
self.key_sound = pg.mixer.Sound(os.path.join(self.snd_dir, KEY))
self.decision_sound = pg.mixer.Sound(
os.path.join(self.snd_dir, DECISION))
self.fail_sound = pg.mixer.Sound(os.path.join(self.snd_dir, FAIL))
self.good_sound = pg.mixer.Sound(os.path.join(self.snd_dir, APPLAUSE))
self.success_sound = pg.mixer.Sound(os.path.join(
self.snd_dir, SUCCESS))
self.exit_sound = pg.mixer.Sound(os.path.join(self.snd_dir, EXIT))
self.blink_sound = pg.mixer.Sound(os.path.join(self.snd_dir, BLINK))
self.count_sound = pg.mixer.Sound(os.path.join(self.snd_dir, COUNT))
self.ready_sound = pg.mixer.Sound(os.path.join(self.snd_dir, READY))
def load_blinkData(self):
with open(PATTERN_DATA_PATH, "r", encoding='UTF-8') as data_file:
data_lines = data_file.read().split('\n')
for data_line in data_lines:
if data_line[0] == '0':
time_list = data_line.split(':')
elif data_line[0] == 'E':
last_line = data_line.split('-')
time_list = last_line[1].split(':')
timing = int(time_list[0]) * 60000 + int(
time_list[1]) * 1000 + int(time_list[2]) * 10
self.blink_data.append(timing)
self.blink_dataLen = len(self.blink_data)
def draw(self):
#game loop - draw
frame = self.vs.read()
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame = cv2.rotate(frame, cv2.ROTATE_90_COUNTERCLOCKWISE)
frame = cv2.flip(frame, flipCode=0)
frame = pg.surfarray.make_surface(frame)
self.screen.blit(frame, (0, 0))
if self.practice == True:
pg.draw.circle(self.screen, YELLOW, (WIDTH / 2, HEIGHT / 2),
WIDTH * 0.01)
self.draw_text("PRACTICE MODE", 15, BLACK, WIDTH / 2,
HEIGHT / 5 - 20, WHITE)
self.draw_text("막대가 화면의 중앙에 올 때 눈을 깜빡여주세요.", 22, BLACK, WIDTH / 2,
HEIGHT / 5)
else:
self.draw_text("PRACTICAL MODE", 15, BLACK, WIDTH / 2,
HEIGHT / 5 - 20, WHITE)
self.draw_text("잠금 해제 중...", 22, BLACK, WIDTH / 2, HEIGHT / 5)
if self.game_tick >= self.blink_data[-1]:
self.ending = True
else:
self.create_blink()
self.all_sprites.draw(self.screen)
pg.display.update()
###-------- start_screen 부분
def start_new(self):
self.start_group = pg.sprite.Group()
self.select = Select(self)
self.start_group.add(self.select)
self.start_run()
def start_run(self):
#start loop
pg.mixer.music.load(os.path.join(self.snd_dir, MAIN))
pg.mixer.music.play(loops=-1)
self.start_playing = True
while self.start_playing:
self.clock.tick(FPS)
self.start_events()
self.start_update()
self.start_draw()
pg.mixer.music.fadeout(200)
def start_events(self):
for event in pg.event.get():
if event.type == pg.QUIT:
if self.start_playing:
self.start_playing = False
self.start = False
def start_update(self):
self.start_group.update()
def start_draw(self):
self.screen.blit(self.bg_img, (0, 0))
self.start_group.draw(self.screen)
self.draw_text("눈 깜빡임 패턴으로 잠금 해제하기", 22, WHITE, WIDTH / 2,
HEIGHT / 2 + 25, BLACK)
if self.select.select_number == 0:
self.draw_text('연습', 46, WHITE, 200, 320)
self.draw_text('실전', 36, BLACK, 320, 320)
self.draw_text('EXIT', 36, BLACK, 440, 320)
elif self.select.select_number == 1:
self.draw_text('연습', 36, BLACK, 200, 320)
self.draw_text('실전', 46, WHITE, 320, 320)
self.draw_text('EXIT', 36, BLACK, 440, 320)
elif self.select.select_number == 2:
self.draw_text('연습', 36, BLACK, 200, 320)
self.draw_text('실전', 36, BLACK, 320, 320)
self.draw_text('EXIT', 46, WHITE, 440, 320)
pg.display.update()
def create_blink(self):
if self.blink_index <= self.blink_dataLen - 1 and self.game_tick >= self.blink_data[
self.blink_index]:
obj_b = Blink(self)
self.blink_sign.append(obj_b)
self.all_sprites.add(obj_b)
self.blinks.add(obj_b)
self.blink_index += 1
###---------------end
def fail_screen(self):
# fail 시에 나타낼 스크린
self.background = pg.Surface((WIDTH, HEIGHT)) # 검은 배경
self.background = self.background.convert()
self.background.fill(BLACK)
self.screen.blit(self.background, (0, 0))
self.fail_sound.play()
self.draw_text("잠금을 해제하지 못했습니다.", 48, WHITE, WIDTH / 2, HEIGHT / 3)
self.draw_text("'Z':main screen, 'ESC':QUIT", 22, WHITE, WIDTH / 2,
HEIGHT * 2 / 3)
pg.display.update()
pg.mixer.music.load(os.path.join(self.snd_dir, WAITING))
pg.mixer.music.play(loops=-1)
self.wait_for_key()
pg.mixer.music.fadeout(200)
def good_screen(self):
# 연습 성공 시에 나타낼 스크린
self.background = pg.Surface((WIDTH, HEIGHT)) # 검은 배경
self.background = self.background.convert()
self.background.fill(BLACK)
self.screen.blit(self.background, (0, 0))
self.good_sound.play()
self.draw_text("참 잘했어요.", 48, WHITE, WIDTH / 2, HEIGHT / 3)
self.draw_text("'Z':main screen, 'ESC':QUIT", 22, WHITE, WIDTH / 2,
HEIGHT * 2 / 3)
pg.display.update()
pg.mixer.music.load(os.path.join(self.snd_dir, GOOD))
pg.mixer.music.play(loops=-1)
self.wait_for_key()
pg.mixer.music.fadeout(200)
#클리어시 나타낼 화면
def ending_screen(self):
self.success_sound.play()
pg.mixer.music.load(os.path.join(self.snd_dir, ENDING))
pg.mixer.music.play(loops=-1)
cv2.destroyAllWindows()
self.screen.blit(self.end_img, (0, 0))
self.draw_text("잠금을 해제했습니다.", 30, WHITE, WIDTH / 2, HEIGHT / 3)
self.draw_text("상명대학교 AI SW 공모전", 20, WHITE, WIDTH / 2, HEIGHT * 2 / 3)
self.draw_text("팀: 느낌알조~?", 18, WHITE, WIDTH / 2, HEIGHT * 2 / 3 + 22)
self.draw_text("지도교수:양희경교수님", 18, WHITE, WIDTH / 2,
HEIGHT * 2 / 3 + 42)
pg.display.update()
sleep(8)
pg.mixer.music.fadeout(200)
sleep(2)
pg.quit()
quit()
#화면대기
def wait_for_key(self):
self.ending = False
waiting = True
while waiting:
self.clock.tick(FPS)
for event in pg.event.get():
if event.type == pg.QUIT:
pg.quit()
quit()
elif event.type == pg.KEYDOWN:
if event.key == pg.K_ESCAPE:
self.exit_sound.play()
sleep(0.5)
pg.quit()
quit()
if event.key == pg.K_z:
self.decision_sound.play()
waiting = False
self.start = True
self.playing = False
sleep(0.14)
sleep(0.1)
#화면에 텍스트 처리를 위한 메서드
def draw_text(self, text, size, color, x, y, back=None):
font = pg.font.Font(self.gg_font, size)
text_surface = font.render(text, True, color, back)
text_rect = text_surface.get_rect()
text_rect.midtop = (x, y)
self.screen.blit(text_surface, text_rect)
#render(text, antialias, color, background=None) -> Surface
def draw_sprite(self, pos, spr, alpha=ALPHA_MAX):
spr.set_alpha(alpha)
self.screen.blit(spr, (round(pos[0]), round(pos[1])))
#Intro
def intro(self):
screen_alpha = 10
pg.mixer.music.load(os.path.join(self.snd_dir, INTRO))
pg.mixer.music.play(loops=-1)
self.background = pg.Surface((WIDTH, HEIGHT)) # 리셋
self.background.fill(WHITE)
while (screen_alpha <= ALPHA_MAX):
self.screen.blit(self.background, (0, 0))
if screen_alpha % 2 == 0:
img = self.team_img1
else:
img = self.team_img2
self.draw_sprite((0, 0), img, screen_alpha)
screen_alpha += 1
pg.display.update()
sleep(0.02)
pg.mixer.music.fadeout(400)
sleep(0.5)
#Ready
def ready(self):
cnt = ['3', '2', '1', 'READY?', 'START!']
for i in cnt:
self.background = pg.Surface((WIDTH, HEIGHT)) # 리셋
self.background.fill(BLACK)
self.screen.blit(self.background, (0, 0))
self.draw_text(i, 70, WHITE, WIDTH / 2, HEIGHT / 2 - 35)
pg.display.update()
if i != cnt[-1]:
self.count_sound.play()
sleep(1)
args=(persist_queue, ),
daemon=True)
persistence_worker.start()
stream = VideoStream(0).start()
time.sleep(2)
sort_tracker = Sort(max_age=20, min_hits=1)
if __name__ == "__main__":
dets = np.array([])
labels = np.array([])
probs_max = np.array([])
faces = np.array([])
fps = FPS().start()
while True:
ret, frame = stream.read()
if not ret:
continue
frame = cv2.resize(frame, (1280, 720))
if fps._numFrames % SKIP_FRAME == 0:
rects = detector(frame, 0)
dets = np.array([get_pos_from_rect(rect) for rect in rects])
labels = np.empty(len(rects))
probs_max = np.empty(len(rects))
faces = np.array([get_face_from_frame(det, frame) for det in dets])
frame_processed = preprocess(frame)[..., ::-1]
if len(rects) > 0:
embeddings = []
for i, rect in enumerate(rects):
shape = predictor(frame_processed, rect)
face_descriptor = facerec.compute_face_descriptor(
vs = VideoStream(device=0).start()
time.sleep(0.1)
prev = time.time()
while RUN:
current = time.time()
# camera stream
if vs.check_queue():
delta = current - prev
prev = current
screen.fill([0,0,0])
frame = vs.read()
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
im = frame.copy()
roi = im[x0:x1, y0:y1]
tmp = np.expand_dims(roi, axis=0)
roi_pred = model.predict(tmp)
if MODE == 'argmax':
if n_classes == 1:
roi_pred = roi_pred.squeeze()
roi_softmax = np.stack([1-roi_pred, roi_pred], axis=2)
roi_max = np.argmax(roi_softmax, axis=2)
roi_pred = np.array(roi_max, dtype=np.float32)
elif n_classes > 1:
roi_max = np.argmax(roi_pred.squeeze(), axis=2)
yes = mlutils.yes_or_no("\n Do you want to continue searching for '{}'", term, yes=True)
if not yes:
break
else:
if args.camera:
src = 0
print("\nStart recognising faces in live camera ...")
else:
src = args.video
print("\nStart recognising faces in video {} ...".format(src))
print("\nPlease type 'q' to stop.")
video = VideoStream(src).start()
while video.running():
frame = video.read()
rgb = cv.cvtColor(frame, cv.COLOR_BGR2RGB)
boxes = detect_faces(rgb)
encodings = encode_faces(boxes, rgb)
for (box, encoding) in zip(boxes, encodings):
found_name = match_face(encoding, candidate_encodings, candidate_names, cnt_dict)
if found_name is not None:
mark_face(frame, box, found_name)
cv.imshow("Frame", frame)
key = cv.waitKey(1) & 0xFF
if key == ord("q"):
break
cv.destroyAllWindows()
video.stop()