-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun_v2.py
533 lines (421 loc) · 19.2 KB
/
run_v2.py
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
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
import cv2
import pickle as pkl
import time
import xgboost as xgb
import math
import numpy as np
import mediapipe as mp
from speak import speakText
mp_drawing = mp.solutions.drawing_utils
mp_hands = mp.solutions.hands
width = 640
height = 480
hands = mp_hands.Hands(min_detection_confidence=0.6, min_tracking_confidence=0.75)
modelRight = pkl.load(open('./models/xgboost-model-dynamic-words-16-right-tuned', 'rb'))
modelLeft = pkl.load(open('./models/xgboost-model-dynamic-words-16-left-tuned', 'rb'))
start_time = time.time()
speakWaitTime = 1
labels = {
"0" : "me",
"1" : "you",
"2" : "hello",
"3" : "from",
"4" : "good",
"5" : "how",
"6" : "university",
"7" : "welcome",
"8" : "hope",
"9" : "like",
"10" : "new",
"11" : "people",
"12" : "technology",
"13" : "use",
"14" : "voice",
"15" : "create"
}
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, height)
rightHandFirst = False
isMultiHand = False
# Initialised to True for first frame
addRightFrame = True
addLeftFrame = True
rightKeyFrames = []
leftKeyFrames = []
rightKeyCheckPoints = []
leftKeyCheckPoints = []
rightLabel = ''
rightProb = 0
leftLabel = ''
leftProb = 0
null_12 = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
d_threshold = 0.1
connections = [
(1, 4), (5, 8), (9, 12), (13, 16), (17, 20)
]
r_c_x = 0
r_c_y = 0
l_c_x = 0
l_c_y = 0
r_p_x = 0
r_p_y = 0
l_p_x = 0
l_p_y = 0
def generatePointVectors(points, previousFrames):
vectors = []
prev_origin_x = 0
prev_origin_y = 0
dx = 0
dy = 0
if(len(previousFrames) == 0):
prev_origin_x = 0
prev_origin_y = 0
else:
prev_origin_x = previousFrames[0]
prev_origin_y = previousFrames[1]
origin_x, origin_y = points[0]
origin_x_rounded = round((origin_x), 5)
origin_y_rounded = round((origin_y), 5)
dx = origin_x_rounded - prev_origin_x
dy = origin_y_rounded - prev_origin_y
vectors.append(origin_x_rounded)
vectors.append(origin_y_rounded)
for num, connection in enumerate(connections):
x0, y0 = points[connection[0]]
x1, y1 = points[connection[1]]
x_final = x1 - x0
y_final = y1 - y0
mag = math.sqrt((x_final)**2+(y_final)**2)
x_vector = round((x_final/mag) + dx,5)
y_vector = round((y_final/mag) + dy,5)
vectors.append(x_vector)
vectors.append(y_vector)
return vectors
def generateCheckPoints(points):
checkPoints = []
palm_x, palm_y = points[0]
thumb_x, thumb_y = points[4]
index_x, index_y = points[8]
pinky_x, pinky_y = points[20]
mean_x = round((palm_x + thumb_x + index_x + pinky_x)/4, 5)
mean_y = round((palm_y + thumb_y + index_y + pinky_y)/4, 5)
checkPoints.append(mean_x)
checkPoints.append(mean_y)
return checkPoints
def checkPreviousFrame(currCheckPoints, prevCheckPoints):
current_dx = currCheckPoints[0]
current_dy = currCheckPoints[1]
prev_dx = prevCheckPoints[0]
prev_dy = prevCheckPoints[1]
dx = round(abs(current_dx - prev_dx), 5)
dy = round(abs(current_dy - prev_dy), 5)
if(dx >= d_threshold or dy >= d_threshold):
print("Thresold crossed.")
return True, current_dx, current_dy, prev_dx, prev_dy
else:
return False, current_dx, current_dy, prev_dx, prev_dy
def recalculateFrames(frames):
cycledFrames = []
cycledFrames.extend(frames)
# Current Origin
if(len(frames) > 12):
base_x = cycledFrames[0]
base_y = cycledFrames[1]
secondFrame_dx = cycledFrames[12] - base_x
secondFrame_dy = cycledFrames[13] - base_y
# New Origin
new_base_x = cycledFrames[12]
new_base_y = cycledFrames[13]
if(len(frames) > 24):
thirdFrame_dx = cycledFrames[24] - base_x
thirdFrame_dy = cycledFrames[25] - base_y
# New second frame
new_secondFrame_dx = cycledFrames[24] - new_base_x
new_secondFrame_dy = cycledFrames[25] - new_base_y
if(len(frames) > 36):
fourthFrame_dx = cycledFrames[36] - base_x
fourthFrame_dy = cycledFrames[37] - base_y
# New third frame
new_thirdFrame_dx = cycledFrames[36] - new_base_x
new_thirdFrame_dy = cycledFrames[37] - new_base_y
i = 12
while(i < 48):
# This
if(i >= 14 and i < 24 and len(frames) > 12):
cycledFrames[i] = round((cycledFrames[i] - secondFrame_dx), 5)
cycledFrames[i + 1] = round((cycledFrames[i + 1] - secondFrame_dy), 5)
# This
elif(i >= 26 and i < 36 and len(frames) > 24):
original_keyframe_dx = cycledFrames[i] - thirdFrame_dx
original_keyframe_dy = cycledFrames[i + 1] - thirdFrame_dy
cycledFrames[i] = round(original_keyframe_dx + new_secondFrame_dx, 5)
cycledFrames[i + 1] = round(original_keyframe_dy + new_secondFrame_dy, 5)
# This
elif(i >= 38 and i < 48 and len(frames) > 36):
original_keyframe_dx = cycledFrames[i] - fourthFrame_dx
original_keyframe_dy = cycledFrames[i + 1] - fourthFrame_dy
cycledFrames[i] = round(original_keyframe_dx + new_thirdFrame_dx, 5)
cycledFrames[i + 1] = round(original_keyframe_dy + new_thirdFrame_dy, 5)
i = i + 2
# 0 - 11
# 12 - 23
# 24 - 35
# 36 - 47
# Cycle out
cycledFrames = cycledFrames[12:]
return cycledFrames
def preprocessData(frames):
dataToProcess = []
dataToProcess.extend(frames)
if(len(dataToProcess) != 48):
if(len(dataToProcess) == 12):
dataToProcess.extend(null_12)
dataToProcess.extend(null_12)
dataToProcess.extend(null_12)
elif(len(dataToProcess) == 24):
dataToProcess.extend(null_12)
dataToProcess.extend(null_12)
elif(len(dataToProcess) == 36):
dataToProcess.extend(null_12)
else:
print("Error in preprocessData. Length of dataToProcess: ", len(dataToProcess))
group_0 = []
group_0.extend(dataToProcess[:12])
group_0.extend(null_12)
group_0.extend(null_12)
group_0.extend(null_12)
group_1 = []
group_1.extend(dataToProcess[:24])
group_1.extend(null_12)
group_1.extend(null_12)
group_2 = []
group_2.extend(dataToProcess[:36])
group_2.extend(null_12)
group_3 = []
group_3.extend(dataToProcess[:48])
arr_0 = np.array(group_0)
arr_1 = np.array(group_1)
arr_2 = np.array(group_2)
arr_3 = np.array(group_3)
inputData_0 = xgb.DMatrix(arr_0.data)
inputData_1 = xgb.DMatrix(arr_1.data)
inputData_2 = xgb.DMatrix(arr_2.data)
inputData_3 = xgb.DMatrix(arr_3.data)
# Convert values to DMatrix format
return xgb.DMatrix(arr_0.data), xgb.DMatrix(arr_1.data), xgb.DMatrix(arr_2.data), xgb.DMatrix(arr_3.data)
def classification(inputData_0, inputData_1, inputData_2, inputData_3, model):
prob_list_0 = model.predict(inputData_0)[0]
prob_list_1 = model.predict(inputData_1)[0]
prob_list_2 = model.predict(inputData_2)[0]
prob_list_3 = model.predict(inputData_3)[0]
max_prob_0 = np.amax(prob_list_0)
max_prob_1 = np.amax(prob_list_1)
max_prob_2 = np.amax(prob_list_2)
max_prob_3 = np.amax(prob_list_3)
out_label_0 = labels["{}".format(np.argmax(prob_list_0, axis=0))]
out_label_1 = labels["{}".format(np.argmax(prob_list_1, axis=0))]
out_label_2 = labels["{}".format(np.argmax(prob_list_2, axis=0))]
out_label_3 = labels["{}".format(np.argmax(prob_list_3, axis=0))]
label = out_label_0
prob = max_prob_0
if(prob < max_prob_1 and max_prob_1 > max_prob_2 and max_prob_1 > max_prob_3):
prob = max_prob_1
label = out_label_1
elif(prob < max_prob_2 and max_prob_2 > max_prob_3 and max_prob_2 > max_prob_1):
prob = max_prob_2
label = out_label_2
elif(prob < max_prob_3 and max_prob_3 > max_prob_1 and max_prob_3 > max_prob_2):
prob = max_prob_3
label = out_label_3
return label, prob
def cleanUp(frames, model):
temp_frames = []
temp_frames.extend(frames)
temp_label = ''
temp_prob = 0
if(model is None):
temp_frames = []
return temp_frames, temp_label, temp_prob
while(len(temp_frames) != 0):
temp_frames = recalculateFrames(temp_frames)
if(len(temp_frames) != 0):
# Preprocess
set0, set1, set2, set3 = preprocessData(temp_frames)
# Classify
temp_label, temp_prob = classification(set0, set1, set2, set3, model)
temp_frames = []
return temp_frames, temp_label, temp_prob
while cap.isOpened():
success, image = cap.read()
if not success:
break
# Flip the image horizontally for a later selfie-view display, and convert the BGR image to RGB.
image = cv2.cvtColor(cv2.flip(image, 1), cv2.COLOR_BGR2RGB)
# To improve performance, optionally mark the image as not writeable to pass by reference.
image.flags.writeable = False
results = hands.process(image)
if(results.multi_handedness):
if(len(results.multi_handedness) == 1):
isMultiHand = False
else:
isMultiHand = True
# results.multi_handedness[0] is first detected hand
if(results.multi_handedness[0].classification[0].index == 0): # Index 0 is Left, 1 is Right
rightHandFirst = False
else:
rightHandFirst = True
# Draw the hand annotations on the image.
image.flags.writeable = True
image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
if results.multi_hand_landmarks:
rightHandPoints = []
leftHandPoints = []
rightVectors = []
leftVectors = []
rightCheckPoints = []
leftCheckPoints = []
for hand, hand_landmarks in enumerate(results.multi_hand_landmarks):
if(rightHandFirst): # First hand (0) is Right, Second hand (1) is Left
if(hand == 0):
for idx, landmark in enumerate(hand_landmarks.landmark):
rightHandPoints.append((landmark.x, landmark.y))
else:
for idx, landmark in enumerate(hand_landmarks.landmark):
leftHandPoints.append((landmark.x, landmark.y))
else: # First hand (0) is Left, Second hand (1) is Right
if(hand == 0):
for idx, landmark in enumerate(hand_landmarks.landmark):
leftHandPoints.append((landmark.x, landmark.y))
else:
for idx, landmark in enumerate(hand_landmarks.landmark):
rightHandPoints.append((landmark.x, landmark.y))
if(isMultiHand):
if(hand == 1):
if(len(rightHandPoints) != 0 and len(leftHandPoints) != 0):
rightVectors = generatePointVectors(rightHandPoints, rightKeyFrames)
rightCheckPoints = generateCheckPoints(rightHandPoints)
leftVectors = generatePointVectors(leftHandPoints, leftKeyFrames)
leftCheckPoints = generateCheckPoints(leftHandPoints)
if(len(rightKeyFrames) == 48):
rightKeyFrames = recalculateFrames(rightKeyFrames)
print("Right Frame Cycled:", len(rightKeyFrames))
if(len(leftKeyFrames) == 48):
leftKeyFrames = recalculateFrames(leftKeyFrames)
print("Left Frame Cycled:", len(leftKeyFrames))
if(addRightFrame == True or addLeftFrame == True):
rightKeyFrames.extend(rightVectors)
rightKeyCheckPoints.extend(rightCheckPoints)
leftKeyFrames.extend(leftVectors)
leftKeyCheckPoints.extend(leftCheckPoints)
print("Right Added: ", len(rightKeyFrames), "Left Added: ", len(leftKeyFrames))
# Preprocess
r_set0, r_set1, r_set2, r_set3 = preprocessData(rightKeyFrames)
l_set0, l_set1, l_set2, l_set3 = preprocessData(leftKeyFrames)
# Classify
rightLabel, rightProb = classification(r_set0, r_set1, r_set2, r_set3, modelRight)
leftLabel, leftProb = classification(l_set0, l_set1, l_set2, l_set3, modelLeft)
addRightFrame = False
addLeftFrame = False
else:
if(len(rightKeyCheckPoints) == 0):
rightKeyCheckPoints.extend(rightCheckPoints)
else:
addRightFrame, r_c_x, r_c_y, r_p_x, r_p_y = checkPreviousFrame(rightCheckPoints, rightKeyCheckPoints)
if(len(leftKeyCheckPoints) == 0):
leftKeyCheckPoints.extend(leftCheckPoints)
else:
addLeftFrame, l_c_x, l_c_y, l_p_x, l_p_y = checkPreviousFrame(leftCheckPoints, leftKeyCheckPoints)
if(addRightFrame == True or addLeftFrame == True):
rightKeyCheckPoints = []
leftKeyCheckPoints = []
if(len(rightKeyFrames) == 48):
rightKeyFrames = recalculateFrames(rightKeyFrames)
print("Right Frame Cycled:", len(rightKeyFrames))
if(len(leftKeyFrames) == 48):
leftKeyFrames = recalculateFrames(leftKeyFrames)
print("Left Frame Cycled:", len(leftKeyFrames))
cv2.circle(image, (int(r_c_x * width), int(r_c_y * height)), 3, (255, 0, 0), 2)
cv2.circle(image, (int(l_c_x * width), int(l_c_y * height)), 3, (255, 255, 0), 2)
cv2.circle(image, (int(r_p_x * width), int(r_p_y * height)), 3, (0, 255, 0), 2)
cv2.circle(image, (int(l_p_x * width), int(l_p_y * height)), 3, (0, 255, 255), 2)
else:
if(len(rightHandPoints) != 0):
rightVectors = generatePointVectors(rightHandPoints, rightKeyFrames)
rightCheckPoints = generateCheckPoints(rightHandPoints)
if(addRightFrame == True):
rightKeyFrames.extend(rightVectors)
rightKeyCheckPoints.extend(rightCheckPoints)
print("Right Frame Added: ", len(rightKeyFrames))
# Preprocess
r_set0, r_set1, r_set2, r_set3 = preprocessData(rightKeyFrames)
# Classify
rightLabel, rightProb = classification(r_set0, r_set1, r_set2, r_set3, modelRight)
leftLabel = ''
leftProb = 0
addRightFrame = False
else:
if(len(rightKeyCheckPoints) == 0):
rightKeyCheckPoints.extend(rightCheckPoints)
else:
addRightFrame, r_c_x, r_c_y, r_p_x, r_p_y = checkPreviousFrame(rightCheckPoints, rightKeyCheckPoints)
if(addRightFrame == True):
rightKeyCheckPoints = []
if(len(rightKeyFrames) == 48):
rightKeyFrames = recalculateFrames(rightKeyFrames)
print("Right Frame Cycled:", len(rightKeyFrames))
if(len(leftKeyFrames) != 0):
leftKeyFrames, leftLabel, leftProb = cleanUp(leftKeyFrames, None)
leftLabel = ''
if(len(leftHandPoints) != 0):
leftVectors = generatePointVectors(leftHandPoints, leftKeyFrames)
leftCheckPoints = generateCheckPoints(leftHandPoints)
if(addLeftFrame == True):
leftKeyFrames.extend(leftVectors)
leftKeyCheckPoints.extend(leftCheckPoints)
print("Left Frame Added: ", len(leftKeyFrames))
# Preprocess
l_set0, l_set1, l_set2, l_set3 = preprocessData(leftKeyFrames)
# Classify
leftLabel, leftProb = classification(l_set0, l_set1, l_set2, l_set3, modelLeft)
rightLabel = ''
rightProb = 0
addLeftFrame = False
else:
if(len(leftKeyCheckPoints) == 0):
leftKeyCheckPoints.extend(leftCheckPoints)
else:
addLeftFrame, l_c_x, l_c_y, l_p_x, l_p_y = checkPreviousFrame(leftCheckPoints, leftKeyCheckPoints)
if(addLeftFrame == True):
leftKeyCheckPoints = []
if(len(leftKeyFrames) == 48):
leftKeyFrames = recalculateFrames(leftKeyFrames)
print("Left Frame Cycled:", len(leftKeyFrames))
if(len(rightKeyFrames) != 0):
rightKeyFrames, rightLabel, rightProb = cleanUp(rightKeyFrames, None)
rightLabel = ''
cv2.circle(image, (int(r_c_x * width), int(r_c_y * height)), 3, (255, 0, 0), 2)
cv2.circle(image, (int(l_c_x * width), int(l_c_y * height)), 3, (255, 255, 0), 2)
cv2.circle(image, (int(r_p_x * width), int(r_p_y * height)), 3, (0, 255, 0), 2)
cv2.circle(image, (int(l_p_x * width), int(l_p_y * height)), 3, (0, 255, 255), 2)
mp_drawing.draw_landmarks(image, hand_landmarks, mp_hands.HAND_CONNECTIONS)
else:
if(len(rightKeyFrames) != 0):
rightKeyFrames, rightLabel, rightProb = cleanUp(rightKeyFrames, modelRight)
rightLabel = ''
if(len(leftKeyFrames) != 0):
leftKeyFrames, leftLabel, leftProb = cleanUp(leftKeyFrames, modelLeft)
leftLabel = ''
cv2.putText(image, rightLabel, (width - 200, height - 10), cv2.FONT_HERSHEY_COMPLEX, 1, (0, 0, 255), 2, 1)
cv2.putText(image, str(rightProb), (10, height - 10), cv2.FONT_HERSHEY_COMPLEX, 1, (0, 0, 255), 2, 1)
# Calculate FPS
if (time.time() - start_time) > speakWaitTime :
# Speak
if rightLabel:
speakText(rightLabel)
start_time = time.time()
cv2.imshow('MediaPipe Hands', image)
if cv2.waitKey(5) & 0xFF == 27:
break
hands.close()
cap.release()