def take():
id_ = str(int(time.time()))
path = 'res/img/' + id_ + '.jpg'
camera.take_photo(path)
detect = vision.detect(path)
result = {'id': id_, 'result': detect}
return jsonify(result)
def pilonhunt():
print 'Looking for pilons...'
motor.ctrl(0)
car_dir.home()
pilons = []
#video_dir.maxright()
#time.sleep(2.0)
video_dir.tilt_rel(0)
r = 0.2
while r < 1:
# r = video_dir.pan(100)
video_dir.tilt_rel(0)
video_dir.pan_rel(r)
r = r + 0.2
time.sleep(1.0)
pilons.append(len(vision.detect()))
video_dir.tilt_rel(0.05)
time.sleep(0.5)
pilons.append(len(vision.detect()))
video_dir.home_x()
print pilons
def distance_from_goal(r_goal, c_goal, t_full):
t, rowsl, colsl, rowsr, colsr = t_full;
is_left = False;
if ((r_goal % 2)== 0):
is_left = True;
savestr1 = '~/jenga/current_pics/left_cam_feedback.JPG';
savestr2 = '~/jenga/current_pics/right_cam_feedback.JPG';
command_string1 = "fswebcam -S 40 -d /dev/video4 " + savestr1;
command_string2 = "fswebcam -D 1 -S 40 -d /dev/video3 " + savestr2;
os.system(command_string1);
os.system(command_string2);
cascade_side = cv2.CascadeClassifier('./detectors/sides.xml');
max_size_side = (33, 69);
cascade_end = cv2.CascadeClassifier('./detectors/ends.xml');
max_size_end = (75, 45);
if (is_left):
pic_end = cv2.imread(savestr1,cv2.CV_LOAD_IMAGE_GRAYSCALE);
pic_side = cv2.imread(savestr2, cv2.CV_LOAD_IMAGE_GRAYSCALE);
side_goal = vision.right_tower[2];
r_block = rowsl[r_goal/2];
c_block = colsl[c_goal];
else:
pic_end = cv2.imread(savestr2,cv2.CV_LOAD_IMAGE_GRAYSCALE);
pic_side = cv2.imread(savestr1, cv2.CV_LOAD_IMAGE_GRAYSCALE);
side_goal = vision.left_tower[0];
r_block = rowsr[r_goal/2];
c_block = colsr[c_goal];
pic_end = cv2.GaussianBlur(pic_end, (15,15), 1);
pic_end = cv2.Canny(pic_end, 70, 175);
pic_side = cv2.GaussianBlur(pic_side, (15,15), 1);
pic_side = cv2.Canny(pic_side, 70, 175);
bbox_sides = vision.detect(pic_side, cascade_side, max_size_side);
bbox_ends = vision.detect(pic_end, cascade_end, max_size_end);
side_error = get_side_error(bbox_sides,r_block, c_block);
end_error = get_end_error(bbox_ends, r_block, c_block);
if (is_left):
return (side_error, end_error[0], end_error[1]);
else:
return (end_error[0], side_error, end_error[1]);
def imageCallback(self, data):
try:
image = self.bridge.imgmsg_to_cv2(data, "bgr8")
except CvBridgeError as e:
print(e)
cv2.imshow("IM", image)
cv2.waitKey(1)
x, y, conf, r = vision.detect(image)
if conf > self.confidence_thres:
ori, angle = vision.calculateOrientation(x, image.shape[1], ZED_FOV_H)
dist = vision.calcDist(2*r, image.shape[1], ZED_FOV_H)
self.notfound = 0
# Angle heading
self.foundCallback(ori, angle, dist)
else:
self.notfound += 1
self.notfoundCallback(self.notfound)
for c_str in ["y", "b", "g"]:
if c_str == "y" : logging.debug("START START CAPTURING YELLOW MANGOS")
elif c_str == "b" : logging.debug("START START CAPTURING BROWN MANGOS")
elif c_str == "g" : logging.debug("START START CAPTURING GREEN MANGOS")
else: break
i = 0
while i < n:
left_raw, right_raw = read_video(cap)
left_dst = cv2.undistort(left_raw, mtx, dist, None, newcameramtx)
right_dst = cv2.undistort(right_raw, mtx, dist, None, newcameramtx)
cv2.imshow('left', left_dst)
cv2.imshow('right', right_dst)
left_mango = vision.detect(left_dst, net)
right_mango = vision.detect(right_dst, net)
if left_mango != None and right_mango != None:
left_i = 2*i
right_i = 2*i + 1
left_path = "mango_color/" + c_str + str(left_i) + ".png"
cv2.imwrite(left_path, left_dst)
logging.debug("WRITE AT " + left_path)
right_path ="mango_color/" + c_str + str(right_i) + ".png"
cv2.imwrite(right_path, right_dst)
logging.debug("WRITE AT " + right_path)
i += 1
cv2.waitKey(100)
if c_str != "g": input("PRESS ENTER TO CONTINUE")
import glob
import vision
import cv2
for imagePath in glob.glob("test_images/*.JPEG"):
image = cv2.imread(imagePath)
x, y, conf, r = vision.detect(image)
vision.draw(image, x, y, r)
def react(id_):
path = 'res/img/' + id_ + '_react.jpg'
camera.take_photo(path)
detect = vision.detect(path)
result = {'id': id_, 'result': detect}
return jsonify(result)
img_counter = 0
while True:
imgResp = urllib.request.urlopen(url)
imgNp = np.array(bytearray(imgResp.read()), dtype=np.uint8)
img = cv2.imdecode(imgNp, -1)
# put the image on screen
img = cv2.resize(img, (540, 540))
cv2.imshow('IPWebcam', img)
#To give the processor some less stress
#time.sleep(0.1)
k = cv2.waitKey(1)
if k % 256 == 27:
# ESC pressed
print("Escape hit, closing...")
break
elif k % 256 == 32:
# SPACE pressed
img_name = "test{}.png".format(img_counter)
cv2.imwrite(img_name, img)
print("{} written!".format(img_name))
detect(img_counter)
img_counter += 1
cv2.destroyAllWindows()
print('READY TO HARVEST')
input('PRESS ENTER TO START')
print("START HARVESTING")
x = x_min
dir_x = 1
loop_end = False
x = x_min
y = y_min
## drive x as primary axis
for x in range(x_min, x_max, x_short):
print('position', x, y)
lframe, rframe = get_frames(vcap)
raw_mangos = vision.detect(rframe, net)
mango = most_left(raw_mangos, x, y)
x_mango, y_mango = approx_position(mango, x, y)
if abs(x_mango - x) <= x_error:
color = get_color(rframe)
#check mango's color
if color == c.yellow() or color == c.brown():
## save point
saved_x = x
## drive y as secondary axis
x, y = go_to(x, y, x, y_mango)
arm.forward()