-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathindex.html
259 lines (202 loc) · 6.84 KB
/
index.html
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
<html>
<meta charset="utf-8"/>
<head>
<script src = "https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]"> </script>
<script src="https://cdn.jsdelivr.net/npm/[email protected]/build/dat.gui.js"></script>
<script src="https://unpkg.com/@tensorflow-models/posenet"></script>
<script src="demo_util.js"></script>
<script src="https://cdn.webrtc-experiment.com/RecordRTC/Whammy.js"></script>
<script src="https://cdn.webrtc-experiment.com/RecordRTC/CanvasRecorder.js"></script>
<script src = "https://cdn.webrtc-experiment.com/screenshot.js"></script>
<script src="https://cdn.webrtc-experiment.com/RecordRTC.js"></script>
</head>
<body>
<div id="main">
<video id="video" src= "vid1.mp4" crossOrigin = "Anonymous" type="video/mp4">
</video>
<br>
<canvas id="output" </canvas>
<br>
</div>
<div>
<button type="button" id="recording">Start Recording</button>
<button type="button" id="play" disabled>Play</button>
<button type="button" id="download" disabled>Download</button>
</div>
<script type="module">
function isAndroid() {
return /Android/i.test(navigator.userAgent);
}
function isiOS() {
return /iPhone|iPad|iPod/i.test(navigator.userAgent);
}
function isMobile() {
return isAndroid() || isiOS();
}
const videoWidth = 600;
const videoHeight = 500;
async function setupCamera() {
const video = document.getElementById('video');
video.width = videoWidth;
video.height = videoHeight;
video.play();
return new Promise((resolve) => {
resolve(video);
});
}
async function loadVideo() {
const video = await setupCamera();
return video;
}
const guiState = {
algorithm: 'multi-pose',
input: {
mobileNetArchitecture: 0.50,
outputStride: 16,
imageScaleFactor: 0.5,
},
singlePoseDetection: {
minPoseConfidence: 0.1,
minPartConfidence: 0.5,
},
multiPoseDetection: {
maxPoseDetections: 3,
minPoseConfidence: 0.5,
minPartConfidence: 0.5,
nmsRadius: 30.0,
},
output: {
showVideo: true,
showSkeleton: true,
showPoints: true,
},
net: null,
};
function setupGui( cameras, net) {
guiState.net = net;
const gui = new dat.GUI({width: 300});
const algorithmController =
gui.add(guiState, 'algorithm', ['single-pose', 'multi-pose']);
let input = gui.addFolder('Input');
// Architecture: there are a few PoseNet models varying in size and
// accuracy. 1.01 is the largest, but will be the slowest. 0.50 is the
// fastest, but least accurate.
const architectureController = input.add(
guiState.input, 'mobileNetArchitecture',
['1.01', '1.00', '0.75', '0.50']);
input.add(guiState.input, 'outputStride', [8, 16, 32]);
// Image scale factor: What to scale the image by before feeding it through
// the network.
input.add(guiState.input, 'imageScaleFactor').min(0.2).max(1.0);
input.open();
let single = gui.addFolder('Single Pose Detection');
single.add(guiState.singlePoseDetection, 'minPoseConfidence', 0.0, 1.0);
single.add(guiState.singlePoseDetection, 'minPartConfidence', 0.0, 1.0);
let multi = gui.addFolder('Multi Pose Detection');
multi.add(guiState.multiPoseDetection, 'maxPoseDetections')
.min(1)
.max(20)
.step(1);
multi.add(guiState.multiPoseDetection, 'minPoseConfidence', 0.0, 1.0);
multi.add(guiState.multiPoseDetection, 'minPartConfidence', 0.0, 1.0);
// nms Radius: controls the minimum distance between poses that are returned
// defaults to 20, which is probably fine for most use cases
multi.add(guiState.multiPoseDetection, 'nmsRadius').min(0.0).max(40.0);
multi.open();
let output = gui.addFolder('Output');
output.add(guiState.output, 'showVideo');
output.add(guiState.output, 'showSkeleton');
output.add(guiState.output, 'showPoints');
output.open();
architectureController.onChange(function(architecture) {
guiState.changeToArchitecture = architecture;
});
algorithmController.onChange(function(value) {
switch (guiState.algorithm) {
case 'single-pose':
multi.close();
single.open();
break;
case 'multi-pose':
single.close();
multi.open();
break;
}
});
}
function detectPoseInRealTime(video, net) {
const canvas = document.getElementById('output');
const ctx = canvas.getContext('2d');
// since images are being fed from a webcam
const flipHorizontal = true;
canvas.width = videoWidth;
canvas.height = videoHeight;
async function poseDetectionFrame() {
if (guiState.changeToArchitecture) {
// Important to purge variables and free up GPU memory
guiState.net.dispose();
// Load the PoseNet model weights for either the 0.50, 0.75, 1.00, or 1.01
// version
guiState.net = await posenet.load(+guiState.changeToArchitecture);
guiState.changeToArchitecture = null;
}
const imageScaleFactor = guiState.input.imageScaleFactor;
const outputStride = +guiState.input.outputStride;
let poses = [];
let minPoseConfidence;
let minPartConfidence;
switch (guiState.algorithm) {
case 'single-pose':
const pose = await guiState.net.estimateSinglePose(
video, imageScaleFactor, flipHorizontal, outputStride);
poses.push(pose);
minPoseConfidence = +guiState.singlePoseDetection.minPoseConfidence;
minPartConfidence = +guiState.singlePoseDetection.minPartConfidence;
break;
case 'multi-pose':
poses = await guiState.net.estimateMultiplePoses(
video, imageScaleFactor, flipHorizontal, outputStride,
guiState.multiPoseDetection.maxPoseDetections,
guiState.multiPoseDetection.minPartConfidence,
guiState.multiPoseDetection.nmsRadius);
minPoseConfidence = +guiState.multiPoseDetection.minPoseConfidence;
minPartConfidence = +guiState.multiPoseDetection.minPartConfidence;
break;
}
ctx.clearRect(0, 0, videoWidth, videoHeight);
if (guiState.output.showVideo) {
ctx.save();
ctx.scale(-1, 1);
ctx.translate(-videoWidth, 0);
ctx.drawImage(video, 0, 0, videoWidth, videoHeight);
ctx.restore();
}
poses.forEach(({score, keypoints}) => {
if (score >= minPoseConfidence) {
if (guiState.output.showPoints) {
drawKeypoints(keypoints, minPartConfidence, ctx);
}
if (guiState.output.showSkeleton) {
drawSkeleton(keypoints, minPartConfidence, ctx);
}
}
});
requestAnimationFrame(poseDetectionFrame);
}
poseDetectionFrame();
}
export async function bindPage() {
const net = await posenet.load(0.75);
let video;
video = await loadVideo();
setupGui([], net);
detectPoseInRealTime(video, net);
}
const button = document.querySelector('button#recording');
button.addEventListener('click', function(e) {
bindPage();
});
</script>
<script src="canvasRecord.js"></script>
</body>
</html>