Spaces:
Sleeping
Sleeping
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +134 -35
src/streamlit_app.py
CHANGED
@@ -2,39 +2,138 @@ import altair as alt
|
|
2 |
import numpy as np
|
3 |
import pandas as pd
|
4 |
import streamlit as st
|
|
|
|
|
|
|
|
|
|
|
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 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
import numpy as np
|
3 |
import pandas as pd
|
4 |
import streamlit as st
|
5 |
+
import cv2
|
6 |
+
import mediapipe as mp
|
7 |
+
import time
|
8 |
+
import os
|
9 |
+
from mediapipe.python.solutions import hands
|
10 |
|
11 |
+
st.set_page_config(page_title="🖐️ Hand Tracking Demo", layout="wide")
|
12 |
+
|
13 |
+
|
14 |
+
# Define constants and helper functions
|
15 |
+
HAND_CONNECTIONS = hands.HAND_CONNECTIONS
|
16 |
+
|
17 |
+
def draw_hand_landmarks(image, hand_landmarks):
|
18 |
+
h, w, _ = image.shape
|
19 |
+
|
20 |
+
# Draw landmarks and connections
|
21 |
+
# Manually draw landmarks as circles
|
22 |
+
for idx, landmark in enumerate(hand_landmarks):
|
23 |
+
cx, cy = int(landmark.x * w), int(landmark.y * h)
|
24 |
+
cv2.circle(image, (cx, cy), 5, (0, 255, 0), -1) # green dots
|
25 |
+
|
26 |
+
# Draw connections (using standard hand skeleton connections)
|
27 |
+
# Define connections between landmarks as per Mediapipe Hands
|
28 |
+
connections = HAND_CONNECTIONS
|
29 |
+
for connection in connections:
|
30 |
+
start_idx, end_idx = connection
|
31 |
+
start = hand_landmarks[start_idx]
|
32 |
+
end = hand_landmarks[end_idx]
|
33 |
+
start_point = (int(start.x * w), int(start.y * h))
|
34 |
+
end_point = (int(end.x * w), int(end.y * h))
|
35 |
+
cv2.line(image, start_point, end_point, (255, 0, 0), 2) # blue lines
|
36 |
+
|
37 |
+
|
38 |
+
BaseOptions = mp.tasks.BaseOptions
|
39 |
+
HandLandmarker = mp.tasks.vision.HandLandmarker
|
40 |
+
HandLandmarkerOptions = mp.tasks.vision.HandLandmarkerOptions
|
41 |
+
VisionRunningMode = mp.tasks.vision.RunningMode
|
42 |
+
model_path = "./models/hand_landmarker.task"
|
43 |
+
options = HandLandmarkerOptions(
|
44 |
+
base_options=BaseOptions(model_asset_path=model_path),
|
45 |
+
running_mode=VisionRunningMode.IMAGE,
|
46 |
+
num_hands=2,
|
47 |
+
)
|
48 |
+
landmarker = HandLandmarker.create_from_options(options)
|
49 |
+
|
50 |
+
|
51 |
+
# Set up Streamlit interface
|
52 |
+
st.title("🖐️ Hand Tracking Demo")
|
53 |
+
# Add a stop button
|
54 |
+
stop_button = st.button("Stop")
|
55 |
+
frame_placeholder = st.empty()
|
56 |
+
finger_count_text = st.empty()
|
57 |
+
# Initialize webcam
|
58 |
+
cap = cv2.VideoCapture(0)
|
59 |
+
# Initialize finger count
|
60 |
+
current_finger_count = 0
|
61 |
+
|
62 |
+
# Initialize MediaPipe Hands
|
63 |
+
with hands.Hands(
|
64 |
+
max_num_hands=2, min_detection_confidence=0.5, min_tracking_confidence=0.5
|
65 |
+
) as hands_model:
|
66 |
+
while not stop_button:
|
67 |
+
ret, frame = cap.read()
|
68 |
+
if not ret:
|
69 |
+
st.error("Failed to capture video from webcam")
|
70 |
+
break
|
71 |
+
frame = cv2.flip(frame, 1)
|
72 |
+
frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
|
73 |
+
mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=frame_rgb)
|
74 |
+
result = landmarker.detect(mp_image)
|
75 |
+
# Reset finger count for each frame
|
76 |
+
current_finger_count = 0
|
77 |
+
if result.hand_landmarks:
|
78 |
+
for hand_landmarks in result.hand_landmarks:
|
79 |
+
# Draw landmarks & connections
|
80 |
+
draw_hand_landmarks(frame, hand_landmarks)
|
81 |
+
# Calculate finger count for this hand
|
82 |
+
hand_finger_count = 0
|
83 |
+
if hand_landmarks[4].y < hand_landmarks[3].y: # Thumb
|
84 |
+
hand_finger_count += 1
|
85 |
+
for i in [8, 12, 16, 20]: # Index, middle, ring, pinky
|
86 |
+
if hand_landmarks[i].y < hand_landmarks[i - 2].y:
|
87 |
+
hand_finger_count += 1
|
88 |
+
# Add this hand's fingers to the total count
|
89 |
+
current_finger_count += hand_finger_count
|
90 |
+
# Display finger count
|
91 |
+
finger_count_text.markdown(f"### Fingers detected: {current_finger_count}")
|
92 |
+
# Convert BGR to RGB for displaying in Streamlit
|
93 |
+
frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
|
94 |
+
frame_placeholder.image(frame_rgb, channels="RGB")
|
95 |
+
# Add a small delay to simulate real-time processing
|
96 |
+
time.sleep(0.05)
|
97 |
+
# Rerun to check if stop button was pressed
|
98 |
+
if stop_button:
|
99 |
+
break
|
100 |
+
# Release resources
|
101 |
+
cap.release()
|
102 |
+
st.success("Camera released. Application stopped.")
|
103 |
+
|
104 |
+
|
105 |
+
# """
|
106 |
+
# # Welcome to Streamlit!
|
107 |
+
|
108 |
+
# Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:.
|
109 |
+
# If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
|
110 |
+
# forums](https://discuss.streamlit.io).
|
111 |
+
|
112 |
+
# In the meantime, below is an example of what you can do with just a few lines of code:
|
113 |
+
# """
|
114 |
+
|
115 |
+
# num_points = st.slider("Number of points in spiral", 1, 10000, 1100)
|
116 |
+
# num_turns = st.slider("Number of turns in spiral", 1, 300, 31)
|
117 |
+
|
118 |
+
# indices = np.linspace(0, 1, num_points)
|
119 |
+
# theta = 2 * np.pi * num_turns * indices
|
120 |
+
# radius = indices
|
121 |
+
|
122 |
+
# x = radius * np.cos(theta)
|
123 |
+
# y = radius * np.sin(theta)
|
124 |
+
|
125 |
+
# df = pd.DataFrame({
|
126 |
+
# "x": x,
|
127 |
+
# "y": y,
|
128 |
+
# "idx": indices,
|
129 |
+
# "rand": np.random.randn(num_points),
|
130 |
+
# })
|
131 |
+
|
132 |
+
# st.altair_chart(alt.Chart(df, height=700, width=700)
|
133 |
+
# .mark_point(filled=True)
|
134 |
+
# .encode(
|
135 |
+
# x=alt.X("x", axis=None),
|
136 |
+
# y=alt.Y("y", axis=None),
|
137 |
+
# color=alt.Color("idx", legend=None, scale=alt.Scale()),
|
138 |
+
# size=alt.Size("rand", legend=None, scale=alt.Scale(range=[1, 150])),
|
139 |
+
# ))
|