SignSync/sign_detection.py at main · Tanu1313/SignSync · 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
69
70
71
72
73
74
75
76
77
78
79
80
import pickle
import cv2
import mediapipe as mp
import numpy as np

# Load trained model
model_dict = pickle.load(open('model.p', 'rb'))
model = model_dict['model']

# Initialize MediaPipe Hands module
mp_hands = mp.solutions.hands
mp_drawing = mp.solutions.drawing_utils
hands = mp_hands.Hands(static_image_mode=False, min_detection_confidence=0.3, min_tracking_confidence=0.3)

# Label mapping (update this if you have different classes)
labels_dict = {0: 'A', 1: 'B', 2: 'L'}

cap = cv2.VideoCapture(0)

while True:
    data_aux = []
    x_ = []
    y_ = []

    ret, frame = cap.read()
    if not ret:
        break
    H, W, _ = frame.shape
    frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

    results = hands.process(frame_rgb)

    if results.multi_hand_landmarks:
        for hand_landmarks in results.multi_hand_landmarks:
            mp_drawing.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)

            for i in range(len(hand_landmarks.landmark)):
                x = hand_landmarks.landmark[i].x
                y = hand_landmarks.landmark[i].y
                x_.append(x)
                y_.append(y)

            # Normalize landmarks by subtracting the min values (as you did in training)
            for i in range(len(hand_landmarks.landmark)):
                x = hand_landmarks.landmark[i].x
                y = hand_landmarks.landmark[i].y
                data_aux.append(x - min(x_))
                data_aux.append(y - min(y_))

        # Debug: Print length of features
        print(f"Extracted features length: {len(data_aux)}")  # Should be 84 (42 landmarks * 2 for x and y)

        # Ensure data has the correct length before prediction
        if len(data_aux) == 84:  # The model expects 84 features (42 landmarks * 2)
            prediction = model.predict([np.asarray(data_aux)])
            predicted_character = labels_dict[int(prediction[0])]
            print(f"Predicted character: {predicted_character}")

            # Convert normalized coordinates back to pixel values
            x1, y1 = int(min(x_) * W) - 10, int(min(y_) * H) - 10
            x2, y2 = int(max(x_) * W) - 10, int(max(y_) * H) - 10

            cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 0), 4)
            cv2.putText(frame, predicted_character, (x1, y1 - 10),
                        cv2.FONT_HERSHEY_SIMPLEX, 1.3, (0, 0, 0), 3)
        else:
            print("Feature extraction error: Expected 84 features, but got:", len(data_aux))

    else:
        print("No hands detected")

    # Display the result
    cv2.imshow('frame', frame)

    # Exit the loop if 'q' is pressed
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()