-
Notifications
You must be signed in to change notification settings - Fork 21
/
drone.py
300 lines (243 loc) · 10.7 KB
/
drone.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
import os
import libardrone.libardrone as libardrone
import time
from threading import Thread, Lock, Condition
import cv2
import numpy
import keras
from YOLO import SimpleNet, convert_yolo_detections, do_nms_sort
from actuators import Actuator
from utils.TinyYoloNet import ReadTinyYOLONetWeights
class YOLODrone(object):
def __init__(self, manual=True):
self.key = None
self.stop = False
self.mutex = None
self.manuel = manual
self.PID = None
self.boxes = None
self.condition = Condition()
self.update = False
self.contours = None
self.boxes_update = False
self.image = None
self.labels = ["aeroplane", "bicycle", "bird", "boat", "bottle", "bus", "car", "cat", "chair", "cow", "diningtable",
"dog", "horse", "motorbike", "person", "pottedplant", "sheep", "sofa", "train", "tvmonitor"]
yoloNet = ReadTinyYOLONetWeights(os.path.join(os.getcwd(), 'weights/yolo-tiny.weights'))
# reshape weights in every layer
for i in range(yoloNet.layer_number):
l = yoloNet.layers[i]
if (l.type == 'CONVOLUTIONAL'):
weight_array = l.weights
n = weight_array.shape[0]
weight_array = weight_array.reshape((n // (l.size * l.size), (l.size * l.size)))[:, ::-1].reshape((n,))
weight_array = numpy.reshape(weight_array, [l.n, l.c, l.size, l.size])
l.weights = weight_array
if (l.type == 'CONNECTED'):
weight_array = l.weights
weight_array = numpy.reshape(weight_array, [l.input_size, l.output_size])
l.weights = weight_array
self.model = SimpleNet(yoloNet)
sgd = keras.optimizers.SGD(lr=0.1, decay=1e-6, momentum=0.9, nesterov=True)
self.model.compile(optimizer=sgd, loss='categorical_crossentropy')
def start(self):
self.drone = libardrone.ARDrone(True)
self.drone.reset()
if self.manuel:
try:
self.mutex = Lock()
t1 = Thread(target=self.getKeyInput, args=())
t2 = Thread(target=self.getVideoStream, args=())
t3 = Thread(target=self.getBoundingBoxes, args=())
t1.start()
t2.start()
t3.start()
t1.join()
t2.join()
t3.join()
except:
print "Error: unable to start thread"
else:
try:
self.mutex = Lock()
t1 = Thread(target=self.autonomousFlight, args=(448, 448, 98, 0.1, self.labels,))
t2 = Thread(target=self.getVideoStream, args=())
t3 = Thread(target=self.getBoundingBoxes, args=())
t1.start()
t2.start()
t3.start()
t1.join()
t2.join()
t3.join()
except:
print "Error: unable to start thread"
print("Shutting down...")
cv2.destroyAllWindows()
self.drone.land()
time.sleep(0.1)
self.drone.halt()
print("Ok.")
def getKeyInput(self):
while not self.stop: # while 'bedingung true'
time.sleep(0.1)
if self.key == "t": # if 'bedingung true'
self.drone.takeoff()
elif self.key == " ":
self.drone.land()
elif self.key == "0":
self.drone.hover()
elif self.key == "w":
self.drone.move_forward()
elif self.key == "s":
self.drone.move_backward()
elif self.key == "a":
self.drone.move_left()
elif self.key == "d":
self.drone.move_right()
elif self.key == "q":
self.drone.turn_left()
elif self.key == "e":
self.drone.turn_right()
elif self.key == "8":
self.drone.move_up()
elif self.key == "2":
self.drone.move_down()
elif self.key == "c":
self.stop = True
else:
self.drone.hover()
if self.key != " ":
self.key = ""
def getVideoStream(self, img_width=448, img_height=448):
while not self.stop:
img = self.image
if img != None:
nav_data = self.drone.get_navdata()
nav_data = nav_data[0]
font = cv2.FONT_HERSHEY_SIMPLEX
font_size = 0.5
cv2.putText(img, 'Altitude: %.0f' % nav_data['altitude'], (5, 15), font, font_size, (255, 255, 255))
cv2.putText(img, 'Battery: %.0f%%' % nav_data['battery'], (5, 30), font, font_size, (255, 255, 255))
cv2.drawContours(img, self.contours, -1, (0, 255, 0), 3)
thresh = 0.2
self.mutex.acquire()
if self.boxes_update:
self.boxes_update = False
for b in self.boxes:
max_class = numpy.argmax(b.probs)
prob = b.probs[max_class]
if (prob > thresh and self.labels[max_class] == "person"):
left = (b.x - b.w / 2.) * img_width
right = (b.x + b.w / 2.) * img_width
top = (b.y - b.h / 2.) * img_height
bot = (b.y + b.h / 2.) * img_height
cv2.rectangle(img, (int(left), int(top)), (int(right), int(bot)), (0, 0, 255), 3)
self.mutex.release()
cv2.imshow('frame', img)
l = cv2.waitKey(150)
if l < 0:
self.key = ""
else:
self.key = chr(l)
if self.key == "c":
self.stop = True
def variance_of_laplacian(self, image):
# compute the Laplacian of the image and then return the focus
# measure, which is simply the variance of the Laplacian
return cv2.Laplacian(image, cv2.CV_64F).var()
def getBoundingBoxes(self):
newest = time.time()
while not self.stop:
try:
pixelarray = self.drone.get_image()
pixelarray = cv2.cvtColor(pixelarray, cv2.COLOR_BGR2RGB)
# Check for Blurry
gray = cv2.cvtColor(pixelarray, cv2.COLOR_RGB2GRAY)
fm = self.variance_of_laplacian(gray)
if fm < 10:
continue
if pixelarray != None:
# ima = pixelarray[120:540]
ima = cv2.resize(pixelarray, (448, 448))
image = cv2.cvtColor(ima, cv2.COLOR_RGB2BGR)
image = numpy.rollaxis(image, 2, 0)
image = image / 255.0
image = image * 2.0 - 1.0
image = numpy.expand_dims(image, axis=0)
out = self.model.predict(image)
predictions = out[0]
boxes = convert_yolo_detections(predictions)
self.mutex.acquire()
self.boxes = do_nms_sort(boxes, 98)
self.image = ima
self.update = True
self.mutex.release()
except:
pass
def autonomousFlight(self, img_width, img_height, num, thresh, labels):
actuator = Actuator(self.drone, img_width, img_width * 0.5)
print self.drone.navdata
while not self.stop:
if self.update == True:
self.update = False
hsv = cv2.cvtColor(self.image, cv2.COLOR_BGR2HSV)
image = cv2.medianBlur(hsv, 3)
# Filter by color red
lower_red_1 = numpy.array([15, 150, 150])
upper_red_1 = numpy.array([35, 255, 255])
image = cv2.inRange(image, lower_red_1, upper_red_1)
# Put on median blur to reduce noise
image = cv2.medianBlur(image, 11)
# Find contours and decide if hat is one of them
contours, hierarchy = cv2.findContours(image.copy(), cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
self.contours = contours
boxes = self.boxes
best_prob = -99999
best_box = -1
best_contour = None
self.mutex.acquire()
for i in range(num):
# for each box, find the class with maximum prob
max_class = numpy.argmax(boxes[i].probs)
prob = boxes[i].probs[max_class]
temp = boxes[i].w
boxes[i].w = boxes[i].h
boxes[i].h = temp
if prob > thresh and labels[max_class] == "person":
for contour in contours:
x, y, w, h = cv2.boundingRect(contour)
left = (boxes[i].x - boxes[i].w / 2.) * img_width
right = (boxes[i].x + boxes[i].w / 2.) * img_width
top = (boxes[i].y - boxes[i].h / 2.) * img_height
bot = (boxes[i].y + boxes[i].h / 2.) * img_height
if not (x + w < left or right < x or y + h < top or bot < y):
if best_prob < prob and w > 30:
print "prob found"
best_prob = prob
best_box = i
best_contour = contour
self.boxes_update = True
if best_box < 0:
# print "No Update"
self.mutex.release()
self.drone.at(libardrone.at_pcmd, False, 0, 0, 0, 0)
continue
b = boxes[best_box]
left = (b.x - b.w / 2.) * img_width
right = (b.x + b.w / 2.) * img_width
top = (b.y - b.h / 2.) * img_height
bot = (b.y + b.h / 2.) * img_height
if (left < 0): left = 0;
if (right > img_width - 1): right = img_width - 1;
if (top < 0): top = 0;
if (bot > img_height - 1): bot = img_height - 1;
width = right - left
height = bot - top
x, y, w, h = cv2.boundingRect(best_contour)
actuator.step(right - width/2., width)
self.mutex.release()
def main():
drone = YOLODrone(manual=False)
drone.start()
if __name__ == '__main__':
main()