Desktop-Wanderer/camera_view.py at main · chenlongos/Desktop-Wanderer · 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
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
"""
Camera-only viewer: streams the robot camera with tennis ball detection overlay.
Shows bounding box, diameter (px), and estimated distance (cm).

Usage:
    python camera_view.py
    Then open http://<device-ip>:8080 in a browser.
"""
import sys
import os
sys.path.append(os.path.dirname(__file__))

import time
import cv2
import numpy as np
import threading
from http.server import HTTPServer, BaseHTTPRequestHandler
from socketserver import ThreadingMixIn

from src.setup import init_app, get_fps
from src.robot_setup import init_robot, get_robot
from src.yolov import yolo_infer

# Calibration: distance_cm = CAL_M / diameter_px + CAL_C
CAL_M = 2892.91
CAL_C = 0.27

# --- Tunable heatmap parameters (thread-safe) ---
_params_lock = threading.Lock()
_hue_target = 40       # H value the heatmap peaks at (0-180)
_sat_target = 200      # S value (0-255)
_val_target = 200      # V value (0-255)
_blur_radius = 9       # GaussianBlur kernel size (odd, >=1)
_hue_sigma = 80.0      # Gaussian sigma for cylindrical HSV distance

# --- Minimal MJPEG stream server ---
_latest_frame = None
_frame_lock = threading.Lock()

_PAGE = b"""\
<!DOCTYPE html>
<html>
<head><meta name="viewport" content="width=device-width,initial-scale=1">
<style>
  body{margin:0;background:#111;color:#eee;font-family:monospace;
       display:flex;flex-direction:column;align-items:center;padding:10px}
  img{width:100%;max-width:960px}
  .controls{display:flex;gap:18px;flex-wrap:wrap;justify-content:center;
            padding:8px 0;font-size:13px}
  .controls label{display:flex;align-items:center;gap:6px}
  .controls input[type=range]{width:120px}
  .val{min-width:28px;text-align:right}
</style></head>
<body>
<div class="controls">
  <label>Hue
    <input type="range" id="hue" min="0" max="180" value="40">
    <span class="val" id="hueV">40</span>
  </label>
  <label>Sat
    <input type="range" id="sat" min="0" max="255" value="200">
    <span class="val" id="satV">200</span>
  </label>
  <label>Val
    <input type="range" id="val" min="0" max="255" value="200">
    <span class="val" id="valV">200</span>
  </label>
  <label>Blur
    <input type="range" id="blur" min="1" max="31" step="2" value="9">
    <span class="val" id="blurV">9</span>
  </label>
  <label>Sigma
    <input type="range" id="sigma" min="1" max="120" value="80">
    <span class="val" id="sigmaV">80</span>
  </label>
</div>
<img src="/stream">
<script>
function send(k,v){fetch('/api/params?'+k+'='+v)}
document.getElementById('hue').oninput=function(){
  document.getElementById('hueV').textContent=this.value;send('hue',this.value)};
document.getElementById('sat').oninput=function(){
  document.getElementById('satV').textContent=this.value;send('sat',this.value)};
document.getElementById('val').oninput=function(){
  document.getElementById('valV').textContent=this.value;send('val',this.value)};
document.getElementById('blur').oninput=function(){
  var v=this.value%2===0?+this.value+1:+this.value;this.value=v;
  document.getElementById('blurV').textContent=v;send('blur',v)};
document.getElementById('sigma').oninput=function(){
  document.getElementById('sigmaV').textContent=this.value;send('sigma',this.value)};
</script>
</body>
</html>
"""

class _Handler(BaseHTTPRequestHandler):
    def do_GET(self):
        global _hue_target, _sat_target, _val_target, _blur_radius, _hue_sigma
        if self.path == '/stream':
            self.send_response(200)
            self.send_header('Content-Type', 'multipart/x-mixed-replace; boundary=frame')
            self.end_headers()
            try:
                while True:
                    with _frame_lock:
                        if _latest_frame is None:
                            continue
                        _, jpeg = cv2.imencode('.jpg', _latest_frame)
                    self.wfile.write(b'--frame\r\nContent-Type: image/jpeg\r\n\r\n')
                    self.wfile.write(jpeg.tobytes())
                    self.wfile.write(b'\r\n')
            except BrokenPipeError:
                pass
        elif self.path.startswith('/api/params'):
            from urllib.parse import urlparse, parse_qs
            qs = parse_qs(urlparse(self.path).query)
            with _params_lock:
                if 'hue' in qs:
                    _hue_target = max(0, min(180, int(qs['hue'][0])))
                if 'sat' in qs:
                    _sat_target = max(0, min(255, int(qs['sat'][0])))
                if 'val' in qs:
                    _val_target = max(0, min(255, int(qs['val'][0])))
                if 'blur' in qs:
                    v = max(1, min(31, int(qs['blur'][0])))
                    _blur_radius = v if v % 2 == 1 else v + 1
                if 'sigma' in qs:
                    _hue_sigma = max(1.0, min(120.0, float(qs['sigma'][0])))
            self.send_response(204)
            self.end_headers()
        else:
            self.send_response(200)
            self.send_header('Content-Type', 'text/html')
            self.end_headers()
            self.wfile.write(_PAGE)

    def log_message(self, *_):
        pass

class _ThreadedServer(ThreadingMixIn, HTTPServer):
    daemon_threads = True

def _start_server(port=8080):
    server = _ThreadedServer(('0.0.0.0', port), _Handler)
    threading.Thread(target=server.serve_forever, daemon=True).start()
    print(f"Stream at http://0.0.0.0:{port}")

# --- Main loop ---
def main():
    global _latest_frame

    init_app()
    init_robot()
    robot = get_robot()
    robot.connect()
    _start_server()

    fps = get_fps()
    print("Streaming camera with tennis ball detection. Ctrl+C to stop.")

    try:
        while True:
            t0 = time.perf_counter()

            obs = robot.get_observation()
            frame = obs["front"]

            # --- Cylindrical HSV distance heatmap (computed on clean frame) ---
            with _params_lock:
                target_hue = _hue_target
                target_sat = _sat_target
                target_val = _val_target
                blur_k = _blur_radius
                sigma = _hue_sigma

            hsv_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
            hue = hsv_frame[:, :, 0].astype(np.float32)   # 0-180
            sat = hsv_frame[:, :, 1].astype(np.float32)   # 0-255
            val = hsv_frame[:, :, 2].astype(np.float32)   # 0-255

            # Convert H+S to Cartesian on the color wheel, V as z-axis
            hue_rad = hue * (np.pi / 90.0)          # 0-180 → 0-2π
            target_rad = target_hue * (np.pi / 90.0)

            dx = sat * np.cos(hue_rad) - target_sat * np.cos(target_rad)
            dy = sat * np.sin(hue_rad) - target_sat * np.sin(target_rad)
            dz = val - target_val

            dist = np.sqrt(dx * dx + dy * dy + dz * dz)

            # Gaussian similarity from distance
            similarity = np.exp(-0.5 * (dist / sigma) ** 2)
            heat = (similarity * 255).astype(np.uint8)

            # Spatial blur to suppress per-pixel noise
            heat = cv2.GaussianBlur(heat, (blur_k, blur_k), 0)

            heatmap = cv2.applyColorMap(heat, cv2.COLORMAP_JET)

            # Detect tennis balls
            detections = yolo_infer(frame)

            for box in detections:
                x, y, w, h = box.x, box.y, box.w, box.h
                cx, cy = x + w // 2, y + h // 2
                diameter_px = max(w, h)
                distance_cm = CAL_M / diameter_px + CAL_C

                cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)
                cv2.circle(frame, (cx, cy), 4, (0, 0, 255), -1)
                cv2.putText(frame, f"{diameter_px}px  {distance_cm:.1f}cm",
                            (x, y - 8), cv2.FONT_HERSHEY_SIMPLEX, 0.55, (0, 255, 0), 2)

            # --- Center crosshair + Gaussian-weighted average color ---
            fH, fW = frame.shape[:2]
            mid_x, mid_y = fW // 2, fH // 2
            cross_len = 20

            # Draw crosshair
            cv2.line(frame, (mid_x - cross_len, mid_y), (mid_x + cross_len, mid_y), (255, 255, 255), 1)
            cv2.line(frame, (mid_x, mid_y - cross_len), (mid_x, mid_y + cross_len), (255, 255, 255), 1)

            # Sample a patch around center and compute Gaussian-weighted average.
            # Gaussian weighting emphasizes the very center and smoothly falls off,
            # so edge pixels / background bleed don't skew the reading.
            R = 30  # patch radius
            y1, y2 = max(mid_y - R, 0), min(mid_y + R, fH)
            x1, x2 = max(mid_x - R, 0), min(mid_x + R, fW)
            patch = frame[y1:y2, x1:x2].astype(np.float64)

            # 2D Gaussian kernel (sigma = R/2 keeps ~95% weight inside the patch)
            gy = np.exp(-0.5 * ((np.arange(patch.shape[0]) - patch.shape[0] / 2) / (R / 2)) ** 2)
            gx = np.exp(-0.5 * ((np.arange(patch.shape[1]) - patch.shape[1] / 2) / (R / 2)) ** 2)
            kernel = np.outer(gy, gx)
            kernel /= kernel.sum()

            avg_b = (patch[:, :, 0] * kernel).sum()
            avg_g = (patch[:, :, 1] * kernel).sum()
            avg_r = (patch[:, :, 2] * kernel).sum()
            avg_bgr = (int(avg_b), int(avg_g), int(avg_r))

            # HSV is more useful for tuning color-based detection
            hsv_pixel = cv2.cvtColor(np.uint8([[avg_bgr]]), cv2.COLOR_BGR2HSV)[0][0]

            # Color swatch + HSV readout next to crosshair
            cv2.rectangle(frame, (mid_x + cross_len + 6, mid_y - 12),
                          (mid_x + cross_len + 30, mid_y + 12), avg_bgr, -1)
            cv2.putText(frame, f"H:{hsv_pixel[0]} S:{hsv_pixel[1]} V:{hsv_pixel[2]}",
                        (mid_x + cross_len + 35, mid_y + 5),
                        cv2.FONT_HERSHEY_SIMPLEX, 0.45, (255, 255, 255), 1)

            # Stack original frame and heatmap side-by-side
            combined = np.hstack((frame, heatmap))

            with _frame_lock:
                _latest_frame = combined

            elapsed = time.perf_counter() - t0
            time.sleep(max(1.0 / fps - elapsed, 0))
    except KeyboardInterrupt:
        print("\nStopping.")
    finally:
        robot.disconnect()

if __name__ == '__main__':
    main()