prav2020/Computer_vision

1import cv2
2import mediapipe as mp
3import gradio as gr
4from collections import deque
5import time
6
7# MediaPipe setup
8mp_hands = mp.solutions.hands
9mp_drawing = mp.solutions.drawing_utils
10hands = mp_hands.Hands(max_num_hands=2, min_detection_confidence=0.7)
11
12prev_wrist_positions = deque(maxlen=10)
13last_sign_time = 0
14cooldown = 1.0
15last_sign = None
16
17bot_responses = {
18    "I love you": "Love you too !",
19    "Hi": "Hello! 👋",
20    "Bye": "Goodbye! 👋",
21    "You": "Yes, you!",
22    "Help": "I'm here to help!"
23}
24
25def fingers_up(hand_landmarks):
26    tips_ids = [4, 8, 12, 16, 20]
27    fingers = []
28
29    if hand_landmarks.landmark[4].x < hand_landmarks.landmark[3].x:
30        fingers.append(1)
31    else:
32        fingers.append(0)
33
34    for tip_id in tips_ids[1:]:
35        if hand_landmarks.landmark[tip_id].y < hand_landmarks.landmark[tip_id-2].y:
36            fingers.append(1)
37        else:
38            fingers.append(0)
39    return fingers
40
41def detect_sign(hand_landmarks):
42    global prev_wrist_positions
43    fingers = fingers_up(hand_landmarks)
44    wrist = hand_landmarks.landmark[0]
45
46    prev_wrist_positions.append(wrist.x)
47    movement = max(prev_wrist_positions) - min(prev_wrist_positions)
48
49    if fingers == [1,1,0,0,1]:
50        return "I love you"
51    elif fingers == [1,0,0,0,0]:
52        return "Help"
53    elif fingers == [0,1,0,0,0]:
54        return "You"
55    elif fingers == [1,1,1,1,1] and movement < 0.03:
56        return "Hi"
57    elif fingers == [1,1,1,1,1] and movement >= 0.03:
58        return "Bye"
59    return None
60
61def process_frame(frame):
62    global last_sign, last_sign_time
63    if frame is None:
64        return None
65
66    frame = cv2.flip(frame, 1)
67    rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
68    results = hands.process(rgb_frame)
69
70    current_time = time.time()
71
72    if results.multi_hand_landmarks:
73        for hand_landmarks in results.multi_hand_landmarks:
74            mp_drawing.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)
75            temp_sign = detect_sign(hand_landmarks)
76            if temp_sign and current_time - last_sign_time > cooldown:
77                last_sign = temp_sign
78                last_sign_time = current_time
79                break
80    else:
81        if current_time - last_sign_time > 2:
82            last_sign = None
83
84    display_sign = last_sign if last_sign else "No sign detected"
85    display_bot = bot_responses[last_sign] if last_sign else ""
86
87    cv2.putText(frame, f"Sign: {display_sign}", (10,50), cv2.FONT_HERSHEY_SIMPLEX, 1, (0,255,0), 2)
88    if display_bot:
89        cv2.putText(frame, f"Bot: {display_bot}", (10,100), cv2.FONT_HERSHEY_SIMPLEX, 1, (255,0,0), 2)
90
91    return cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
92
93# Gradio interface
94iface = gr.Interface(
95    fn=process_frame,
96    inputs=gr.Image(source="webcam", streaming=True),
97    outputs="image",
98    live=True,
99    title="ASL Chatbot",
100    description="Detects ASL signs from webcam and responds with a chatbot-like reply."
101)
102
103if __name__ == "__main__":
104    iface.launch()
105