HandSign-Detection/app.py at main · MatheusLevy/HandSign-Detection · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
import cv2
from utils import *
import mediapipe as mp
from configs import *
from landmark_classification import HandLandmarkClassificator

def init_hand_landmarks_model():
    mp_hands = mp.solutions.hands
    mp_hands_landmark_detector = mp_hands.Hands(
        static_image_mode= True,
        max_num_hands= num_hands,
        min_detection_confidence= min_detection_confidence,
        min_tracking_confidence= min_tracking_confidence,
    )
    return mp_hands_landmark_detector

def process_landmarks(frame, landmarks, classificator):
    for hand_landmark, _ in zip(landmarks.multi_hand_landmarks, landmarks.multi_handedness):
        hand_bounding_box = bounding_box_from_landmarks(frame, hand_landmark)
        list_of_landmarks = convert_landmarks_to_list(frame, hand_landmark)
        label_predict = classificator.classify(list_of_landmarks)
        frame = draw_landmarks(frame, list_of_landmarks)
        frame = draw_bounding_box(frame, hand_bounding_box, label_predict)
        cv2.imshow('frame', frame)

def save_landmarks_mode(frame, landmarks, key):
    for hand_landmark, _ in zip(landmarks.multi_hand_landmarks, landmarks.multi_handedness):
        hand_bounding_box = bounding_box_from_landmarks(frame, hand_landmark)
        frame = draw_bounding_box(frame, hand_bounding_box)
        list_of_landmarks = convert_landmarks_to_list(frame, hand_landmark)
        frame = draw_landmarks(frame, list_of_landmarks)
        if is_number_key(key):
            save_land_mark(list_of_landmarks, int(chr(key & 0xFF)), landmark_csv_path)
        cv2.imshow('frame', frame)

def capture_webcam(cap_width, cap_height):
    webcam = init_video_capture_device(cap_height=cap_height, cap_width=cap_width)
    cvFpsCalc = CvFpsCalc(buffer_len=10)
    mp_hands_landmark_detector = init_hand_landmarks_model()
    landmark_classificator= HandLandmarkClassificator()
    global mode

    while(True):
        _, frame = webcam.read()
        input_key = cv2.waitKey(10)
        frame = preprocess_frame(frame, cvFpsCalc, mode)
        landmarks = mp_hands_landmark_detector.process(frame)
        frame = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
        if exist_landmarks(landmarks) and mode ==  "predict":
            process_landmarks(frame, landmarks, landmark_classificator)
        elif exist_landmarks(landmarks) and mode == "save":
            save_landmarks_mode(frame, landmarks, input_key)
        else:
            cv2.imshow('frame', frame)

        if pressed_key(input_key, 'q'):
            break

        if pressed_key(input_key, 's'):
            mode = "save"

        if pressed_key(input_key, 'p'):
            mode= "predict"

    webcam.release()
    cv2.destroyAllWindows()

capture_webcam(frame_width, frame_height)