class Client:
def __init__(self, config: ConfigurationFile):
self.config = config
self.database = Database(self.config.db_file)
self.organizer = Organizer(self.config, self.config.db_file)
self.downloader = Downloader(self.config.db_file, self.organizer,
self.config)
self.tracker = Tracker(self.config.db_file, self.downloader,
self.config.update_period)
self.tracker.start()
self.downloader.start()
self.organizer.start()
def add_tvshow(self, tvshow_id: int):
tvshow_name = showrss.get_name(tvshow_id)
self.database.put_tvshow(TVShow(tvshow_id, tvshow_name))
def remove_tvshow(self, tvshow_id: int):
self.database.remove_tvshow(tvshow_id)
def list_tvshows(self):
return self.database.tvshows()
def list_episodes(self, state: EpisodeState = None):
return self.database.episodes(state)
def download_progress(self):
return self.downloader.downloads()
def exit(self):
self.tracker.stop()
self.downloader.stop()
def main ():
#Introduction
print ("This program graphically depicts the flight of a cannonball. ")
print ()
#Get inputs
a = eval(input("Enter the launch angle in degrees: "))
v = eval(input("Enter the initial velocity in meters per second: "))
h = eval(input("Enter the initial height in meters: "))
#Create tracker
projectile = Projectile(a, v, h)
win = GraphWin(200, 200)
win.setCoords(0.0, 0.0, 25.0, 25.0)
tracker = Tracker(win, projectile)
time = 0.0
while projectile.getY() >= -5:
time += .0005
projectile.update(time)
tracker.update()
#Close window
win.getMouse()
win.close()
def compute_gradient((model, dialog, n_data, )):
print 'here'
return 1
tracker = Tracker(model)
tracker.new_dialog()
last_state = tracker.get_state()
accum_loss_grad = []
for shape in model.shapes:
accum_loss_grad.append(np.zeros(shape, dtype=theano.config.floatX))
total_loss = 0.0
for act in dialog:
act_ndx = model.acts[act]
# Run tracker to get the new state and the true state.
curr_state, true_state = tracker.next(act)
# Compute the loss & gradient of the loss.
val = [model.values[true_state[slot]] for slot in model.slots]
total_loss += model.f_curr_slot_loss(curr_state, val)
for param_loss_grad, accum in zip(model.loss_grads, accum_loss_grad):
accum += 1.0 / n_data * param_loss_grad(last_state, act_ndx, val)
last_state = curr_state
return accum_loss_grad, total_loss
class ObjectTracker(OpenCVVideoFilter):
__gstmetadata__ = (
"Object tracker plugin",
"newelement",
"Description",
"Contact")
def __init__(self):
super(ObjectTracker, self).__init__()
self.tracker = Tracker([self.h, self.w])
self.api = GstVideo.VideoRegionOfInterestMeta.get_info()
def do_transform(self, b, outbuffer):
img = self.gst_to_cv(b)
ts = time.time()
self.tracker.track(ts, img)
tag_list = Gst.TagList.new_empty()
for t in self.tracker.to_log():
blob = t.bestblob()
info = Gst.Structure.new_empty('roi')
info.set_value('id', t.id)
info.set_value('ts', ts)
info.set_value('bbox', blob.bbox )
sample = Gst.Sample.new(b, None, None, info)
logging.debug('new tag with object track info for track %s',str(t.id))
tag_list.add_value(Gst.TagMergeMode.APPEND, Gst.TAG_APPLICATION_DATA, sample)
if tag_list.n_tags()>0:
outbuffer.add_meta(self.api, tag_list)
self.post_message(Gst.Message.new_tag(self, tag_list))
return Gst.FlowReturn.OK
def test_canConvertItemToType(self):
tracker = Tracker()
contents = mock()
timezone = mock()
item = tracker._convertToItem(testType, contents, timezone )
storedItem, storedTimezone = item.contains()
self.assertEqual(storedItem, contents)
self.assertEqual(storedTimezone, timezone)
def iterations_from_file(config, filename):
tracker = Tracker(dbdir=config.tracker.get('local_store_directory'))
f = open(filename)
contents = ''
for l in f: contents += l
f.close()
iterations = tracker.parse_stories(contents)
return iterations
def beer_fitness(population):
for p in population:
#perform simulation
points = []
for i in range(10):
# print p.ann
t = Tracker(p.ann)
points.append(t.run())
p.fitness = sum(points)*1.0/len(points)*1.0
def main ():
#Introduction
print ("This program uses a cannonball to shoot at a target.")
#Create graphics window
win = GraphWin(200, 200)
win.setCoords(0.0, 0.0, 25.0, 25.0)
target = Target(win)
flag = False
targetHit = False
while not flag:
#Get inputs
print ()
a = eval(input("Enter the launch angle in degrees: "))
v = eval(input("Enter the initial velocity in meters per second: "))
h = eval(input("Enter the initial height in meters: "))
#Create tracker
projectile = Projectile(a, v, h)
tracker = Tracker(win, projectile)
time = 0.0
while projectile.getY() >= -5:
time += .0005
projectile.update(time)
tracker.update()
#Calculate if cannonball hit target
points = target.points()
center = tracker.circ.getCenter()
center_x = center.getX()
center_y = center.getY()
radius = tracker.circ.getRadius()
for point in points:
x = point.getX()
y = point.getY()
square_dist = (center_x-x) ** 2 + (center_y-y) ** 2
if square_dist <= radius ** 2:
targetHit = True
if targetHit:
print ("\nYou hit the target!")
flag = True
else:
flag = False
print ("\nTry again!")
#Close window
win.getMouse()
win.close()
def linear_regression(points):
if Tracker.distance(points[0], meanPoint(points)) < 10:
return [0, 1]
num_pts = len(points)
xc = points[:, [0]]
yc = points[:, [1]]
ones = np.ones((num_pts, 1))
X = np.concatenate((ones, xc), axis=1)
X_intermediary = np.dot(X.T, X)
Y_intermediary = np.dot(X.T, yc)
R = np.dot(inv(X_intermediary), Y_intermediary)
k = R[1][0]
last_2 = points[-2:]
first_x = last_2[0][0]
second_x = last_2[1][0]
if first_x < second_x:
return [1, k]
else:
return [-1, -k]
def translate(self, readline, result=None, no_imports=None):
# Tracker to keep track of information as the file is processed
self.tokens = Tokens(self.default_kls)
self.tracker = Tracker(result, self.tokens, self.wrapped_setup)
# Add import stuff at the top of the file
if self.import_tokens and no_imports is not True:
self.tracker.add_tokens(self.import_tokens)
# Looking at all the tokens
with self.tracker.add_phase() as tracker:
for tokenum, value, (_, scol), _, _ in generate_tokens(readline):
self.tracker.next_token(tokenum, value, scol)
# Add attributes to our Describes so that the plugin can handle some nesting issues
# Where we have tests in upper level describes being run in lower level describes
if self.with_describe_attrs:
self.tracker.add_tokens(self.tracker.make_describe_attrs())
# If setups should be wrapped, then do this at the bottom
if self.wrapped_setup:
self.tracker.add_tokens(self.tracker.wrapped_setups())
# Add lines to bottom of file to add __testname__ attributes
self.tracker.add_tokens(self.tracker.make_method_names())
# Return translated list of tokens
return self.tracker.result
def __init__(self, window):
self.gwin = window
self.projectile = Projectile(0, 0, 10)
self.tracker = Tracker(self.gwin, self.projectile)
self.target = Target(self.gwin)
self.hits = False
class Torrent_Downloader():
''' Manages download logic:
- Creation and removal of peers.
- Book keeping of pieces downloaded and in progress.
- Checking completed pieces and writing to file.
'''
def __init__(self, torrent, start_listener_callback):
self.torrent = torrent
self.message_handler = MessageHandler(self.torrent, self)
self.start_listener_callback = start_listener_callback
self.ip = self.get_IP_address()
self.tracker = Tracker(self.torrent.announce, self.torrent.get_params())
self.peers = self.create_peers()
self.io_loop = get_event_loop()
self.index = 0
self.callback_dict = {
'check_piece' : self.torrent.check_piece_callback,
'pieces_changed' : self.pieces_changed_callback,
'start_listener' : self.start_listener_callback,
}
self.pieces_needed = []
def get_IP_address(self):
response = get('http://api.ipify.org?format=json')
ip_object = loads(response.text)
return ip_object["ip"]
def create_peers(self):
peers = []
for p in self.tracker.parse_peer_address():
if p[0] == self.ip:
continue
peers.append(Peer(p[0], p[1], self))
return peers
def pieces_changed_callback(self, peer):
''' Check if connected peer has pieces I need. Send interested message.
Call choose_piece.
If peer has no pieces I need, disconnect and remove from peers list.
'''
self.torrent.update_pieces_needed()
for i in self.torrent.pieces_needed:
if peer.has_pieces[i]:
self.io_loop.create_task(self.message_handler.send_message(peer=peer, message_id=2))
self.choose_piece(peer=peer)
break
else:
self.peers.remove(peer)
def choose_piece(self, peer):
''' Finds the next needed piece, updates self.have and self.pieces_needed.
calls construct_request_payload.
'''
piece_index = self.torrent.pieces_needed[0]
self.torrent.have[piece_index] = True
self.torrent.update_pieces_needed()
self.message_handler.construct_request_payload(peer=peer, piece_index=piece_index)
def main():
file_location, download_dir = get_args()
# parse the torrent file and make into an object
torrentfile = TorrentFile(file_location)
# create bittorrent client (brain behind everything)
client = BitTorrentClient(download_dir, torrentfile)
client.create_files()
# create tracker and add it to the client
tracker = Tracker(client.peer_id, torrentfile)
client.add_tracker(tracker)
# connect to the peers associated to the tracker
tracker.get_peers_and_connect(client, torrentfile)
reactor.run()
def main(ball):
c = Camera()
color = ball
log.info("Ball: " + color)
t = BallTracker(color)
ballpos = None
while ballpos is None:
frame = c.get_frame()
ballpos = t.getBallCoordinates(frame)
ballpos = Tracker.transformCoordstoDecartes(ballpos)
previous_positions = np.array([ballpos])
k = 20
while True:
frame = c.get_frame()
ballpos = t.getBallCoordinates(frame)
log.debug(ballpos)
if ballpos is None:
continue
ballpos = Tracker.transformCoordstoDecartes(ballpos)
previous_positions = np.append(previous_positions, [ballpos], axis=0)
last_k_positions = previous_positions[-k:]
direction_vector = linear_regression(last_k_positions)
direction_vector = [20 * x for x in direction_vector]
p1 = Tracker.transformCoordstoCV(round_point(ballpos))
p2 = Tracker.transformCoordstoCV(round_point(add_points(
ballpos, direction_vector)))
frame = cv2.circle(frame, p1, 20, (0, 0, 0), 2)
cv2.line(frame, p1, p2, (0, 255, 122), 2)
cv2.imshow('frame', frame)
l = cv2.waitKey(5) & 0xFF
if l == 27:
break
c.close()
cv2.destroyAllWindows()
class Main:
def __init__(self, trackerUrl, trackerPass):
self.tracker = Tracker(trackerUrl, trackerPass)
def packetHandler(self, aprsString):
print 'APRS String: %s' % aprsString
packet = APRSPacket()
if packet.parse(aprsString):
print '%s -> %s' % (packet.source, packet.dest)
print 'Report type: %s' % packet.reportType
if packet.hasLocation:
print 'Time: %sZ' % packet.time
print 'Coordinates: %f, %f, Altitude: %d ft' % (packet.latitude, packet.longitude, packet.altitude)
print 'Course: %d, Speed: %d kn, Bearing: %d' % (packet.course, packet.speed, packet.bearing)
print 'Comment: %s' % packet.comment
print 'Uploading to tracker'
self.tracker.track(packet)
print ''
def main():
options, files = getopt.getopt(sys.argv[1:], 'ivt:', ['info', 'verbose', 'threads:'])
for flag, value in options:
if flag == '-v' or flag == '--verbose':
configuration['verbose'] = True
elif flag == '-t' or flag == '--threads':
try:
configuration['threads'] = int(value)
except:
usage()
elif flag == '-i' or flag == '--info':
configuration['info'] = True
else:
usage()
if len(files) != 1:
usage()
try:
torrent = Torrent(files[0])
except:
print 'Impossible to read torrent file/magnet link:', files[0]
exit(1)
if configuration['info'] == True:
torrent.show()
exit(0)
torrent.start()
generate_clientid()
tracker = Tracker(torrent.data['announce'])
tracker.update(torrent)
swarm = Swarm()
swarm.update_peers(tracker)
threads = configuration['threads']
def visualize(self, out_filename="out/training_bs.html", out_filename_pickle="out/training_bs.pickle"):
# Do bootstrap for the confusion table.
n_bs = 1
widgets = [progressbar.Percentage(),
' ', progressbar.Bar(),
' ', progressbar.ETA(),
' ', progressbar.AdaptiveETA()]
bs_progress = progressbar.ProgressBar(widgets=widgets).start()
cts = []
for bs_iter in bs_progress(range(n_bs)):
n_dialogs = len(self.training_dialogs)
dataset = self.training_dialogs
tracker = Tracker(self.model, inv=False)
tracker.simulate(dataset)
cts.append(tracker.out_data['confusion_tables'])
ct = bootstrap.from_all_confusion_tables(cts)
context = {}
context['tracker'] = tracker.out_data
context['bootstrap_ct'] = ct
context['mean_score'] = np.mean([ctt.mean_score for ctt in ct.values()])
context['model'] = self.model
context['training_metrics'] = self.training_metrics
context['training_data'] = self.training_dialogs
env = Environment(loader=FileSystemLoader('tpl'))
env.globals.update(zip=zip)
tpl = env.get_template('training.html')
with open(out_filename, "w") as f_out:
f_out.write(tpl.render(**context))
with open(out_filename_pickle, "w") as f_out:
info = {
'mean_score': float(context['mean_score']),
'losses': [float(x) for x in context['training_metrics']['losses']],
#'simulation': context['tracker']['simulation']
}
f_out.write(json.dumps(info))
def __init__(self, config: ConfigurationFile):
self.config = config
self.database = Database(self.config.db_file)
self.organizer = Organizer(self.config, self.config.db_file)
self.downloader = Downloader(self.config.db_file, self.organizer,
self.config)
self.tracker = Tracker(self.config.db_file, self.downloader,
self.config.update_period)
self.tracker.start()
self.downloader.start()
self.organizer.start()
def __init__(self, lines, line_size, verbose): self.statistics_tracker = Tracker() self.processors = 4 self.verbose = verbose shared_bus = Bus() # Register one cache for each processor self.caches = [] for cache_id in range(self.processors): cache = Cache(cache_id, lines, line_size, shared_bus, self.statistics_tracker, self.verbose) self.caches.append(cache)
def do_trackers(self, callback=None):
assert(len(self.hash) == 20)
# talk to trackers in attempt to get peers
if not self.started(): raise StopIteration
if not self.meta: raise StopIteration
if 'announce-list' in self.meta:
tier1 = self.meta['announce-list'][0]
for url in tier1:
tracker = Tracker.instantiate(url, self.hash)
self.trackers[tracker.get_key()] = tracker
if tracker.can_announce():
result = yield gen.Task( tracker.announce )
logging.info('%s tracker announce got result %s' % ([self.hash], result))
def __init__(self, torrent, start_listener_callback):
self.torrent = torrent
self.message_handler = MessageHandler(self.torrent, self)
self.start_listener_callback = start_listener_callback
self.ip = self.get_IP_address()
self.tracker = Tracker(self.torrent.announce, self.torrent.get_params())
self.peers = self.create_peers()
self.io_loop = get_event_loop()
self.index = 0
self.callback_dict = {
'check_piece' : self.torrent.check_piece_callback,
'pieces_changed' : self.pieces_changed_callback,
'start_listener' : self.start_listener_callback,
}
self.pieces_needed = []
def __init__(self, **kwargs):
"""Initialize the keycode bindings and all controllers, create output dir."""
settings = {
'experimentName' : 'someNameHere',
'outputRootDir' : os.path.normpath(os.path.expanduser("~/kauferdata")),
'keyBindings' : {
'quit' : 'q',
'toggleStimulator' : 't',
'triggerStimulator' : 's'
},
'audio' : {},
'gui' : {},
'stimulator' : { 'activeProtocolName' : 'nucleusAccumbensExample' },
'tracker' : {},
'videoIn' : {},
'videoOut' : {},
'writer' : {},
}
settings.update(kwargs)
trialDir = "{0}_{1}_{2}".format(time.strftime("%y%m%d%H%M%S"),
settings['experimentName'], settings['stimulator']['activeProtocolName'])
settings['outputDataDir'] = os.path.join(settings['outputRootDir'], trialDir)
self.keyBindings = settings['keyBindings']
self.keycodeBindings = {k: getattr(opencvgui.keycodes, v)
for k,v in self.keyBindings.items() }
os.makedirs(os.path.join(settings['outputDataDir'], 'audio'))
self.writeExperimentData(settings)
for x in ['audio', 'videoOut', 'writer']:
kwargs[x] = {} if not kwargs.has_key(x) else kwargs[x]
kwargs[x]['outputDataDir'] = settings['outputDataDir']
# gui and writer both need a reference to the app instance because they
# draw on the state of so many other controllers
self.audio = Audio(**settings['audio'])
self.gui = Gui(self, **settings['gui'])
self.stimulator = Stimulator(**settings['stimulator'])
self.tracker = Tracker(**settings['tracker'])
self.videoIn = VideoIn(**settings['videoIn'])
self.videoOut = VideoOut(**settings['videoOut'])
self.writer = Writer(self, **settings['writer'])
def __init__(self, torr_dir="c:/torrent", data_dir="c:/torrent/files", tracker_ip="10.0.0.1:8000", callback=lambda: ".", session_dir="c:/torrent", db_name="torrent.db", ext=".torrent"):
"""
Intialize class.
Parameters
torr_dir: Location for torrent meta data files
data_dir: Location for data files
tracker_ip: IP:PORT string of tracker
callback: option function which will be called while attempting to seed torrents.
torr_db: Name of anydbm database where we save what we're serving.
"""
self.callback = callback
self.data_dir = data_dir
self.torr_dir = torr_dir
self.ext = ext
self.torr_db = os.path.join(session_dir, db_name)
self.db = None # pointer to our torrent DB
self.my_tracker = Tracker(tracker_ip)
self.session = Session(session_dir, db_name)
self.session.register(self.callback)
class Tokeniser(object):
"""Endpoint for tokenising a file"""
def __init__(self, default_kls='object', with_describe_attrs=True, import_tokens=None):
self.default_kls = default_kls
self.import_tokens = import_tokens
self.with_describe_attrs = with_describe_attrs
########################
### TRANSLATE
########################
def translate(self, readline, result=None, no_imports=None):
# Tracker to keep track of information as the file is processed
self.tokens = Tokens(self.default_kls)
self.tracker = Tracker(result, self.tokens)
# Add import stuff at the top of the file
if self.import_tokens and no_imports is not True:
self.tracker.add_tokens(self.import_tokens)
# Looking at all the tokens
with self.tracker.add_phase() as tracker:
for tokenum, value, (_, scol), _, _ in generate_tokens(readline):
self.tracker.next_token(tokenum, value, scol)
# Add attributes to our Describes so that the plugin can handle some nesting issues
# Where we have tests in upper level describes being run in lower level describes
if self.with_describe_attrs:
self.tracker.add_tokens(self.tracker.make_describe_attrs())
# Add lines to bottom of file to add __testname__ attributes
self.tracker.add_tokens(self.tracker.make_method_names())
# Return translated list of tokens
return self.tracker.result
def __init__(self, *args):
Tracker.__init__(self, *args)
self.l = 400.0/self.video.image_shape()[1]
self.radius = 24.0/self.video.image_shape()[1]
self.renderer = PendulumLayer(self.l, self.radius)
from torrent_file import TorrentFile
import argparse
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Bittorrent client')
parser.add_argument('--tests', action='store_true', help='Run doctests')
parser.add_argument('--url',
default='http://173.230.132.32',
help='URL of tracker')
parser.add_argument('--gen',
action='store_true',
help='Generate new torrent file with random data.')
parser.add_argument('--fname',
metavar='filename',
default='myfile.torrent',
help='File to use for generation or reading.')
args = parser.parse_args()
if args.gen:
# Generate new torrent file
# torrent_file = torrent_file.TorrentFile(info_dict, args.fname, True)
pass
else:
torrent_file = TorrentFile(args.fname)
if args.tests:
import doctest
doctest.testmod()
tracker = Tracker(torrent_file)
tracker.connect()
class RoadEstimation:
def __init__(self):
# specify topic and data type
self.sub_bbox_1 = rospy.Subscriber("camera1/detection/vision_objects",
DetectedObjectArray,
self.bbox_array_callback)
self.sub_bbox_6 = rospy.Subscriber("camera6/detection/vision_objects",
DetectedObjectArray,
self.bbox_array_callback)
self.sub_pcd = rospy.Subscriber("/points_raw", PointCloud2,
self.pcd_callback)
# self.sub_pose = rospy.Subscriber("/current_pose", PoseStamped, self.pose_callback)
# Publisher
self.pub_intersect_markers = rospy.Publisher(
"/vision_objects_position_rviz", MarkerArray, queue_size=10)
self.pub_plane_markers = rospy.Publisher("/estimated_plane_rviz",
MarkerArray,
queue_size=10)
self.pub_plane = rospy.Publisher("/estimated_plane",
Plane,
queue_size=10)
self.pub_pcd_inlier = rospy.Publisher("/points_inlier",
PointCloud2,
queue_size=10)
self.pub_pcd_outlier = rospy.Publisher("/points_outlier",
PointCloud2,
queue_size=10)
self.cam1 = camera_setup_1()
self.cam6 = camera_setup_6()
self.plane = Plane3D(-0.157, 0, 0.988, 1.9)
self.plane_world = None
self.plane_tracker = None
self.sac = RANSAC(Plane3D,
min_sample_num=3,
threshold=0.22,
iteration=200,
method="MSAC")
self.tf_listener = TransformListener()
self.tfmr = Transformer()
self.tf_ros = TransformerROS()
def display_bboxes_in_world(self, cam, bboxes):
vis_array = MarkerArray()
xlist, ylist = [], []
for box_id, bbox in enumerate(bboxes):
d, C = cam.bounding_box_to_ray(bbox)
intersection = self.plane.plane_ray_intersection(d, C)
# print(intersection[0,0],'\t', intersection[1,0],'\t', intersection[2,0])
marker = visualize_marker(intersection,
mkr_id=box_id,
scale=0.5,
frame_id="velodyne",
mkr_color=[0.0, 0.8, 0.0, 1.0])
vis_array.markers.append(marker)
xlist.append(intersection[0, 0])
ylist.append(intersection[1, 0])
self.pub_intersect_markers.publish(vis_array)
# plt.pause(0.001)
def bbox_array_callback(self, msg):
if msg.header.frame_id == "camera1":
cam = self.cam1
elif msg.header.frame_id == "camera6":
cam = self.cam6
else:
rospy.logwarn(
"unrecognized frame id {}, bounding box callback in road estimation fail",
msg.header.frame_id)
return
# rospy.loginfo("camera {:d} message received!!".format(camera.id))
bboxes = []
for obj in msg.objects:
# rospy.loginfo("{}".format(obj.label))
# rospy.loginfo("x:{} y:{} width:{} height:{} angle:{}".format(obj.x, obj.y, obj.width, obj.height, obj.angle))
bbox = BoundingBox(obj.x,
obj.y,
obj.width,
obj.height,
obj.angle,
label=obj.label)
bboxes.append(bbox)
self.display_bboxes_in_world(cam, bboxes)
def estimate_plane(self, pcd):
# threshold_z = [2.0, -0.5]
# pcd_z = clip_pcd_by_distance_plane(pcd, self.plane, threshold_z, in_only=True)
vec1 = np.array([1, 0, 0])
vec2 = np.array([0, 0, 1])
pt1 = np.array([0, 0, 0])
threshold = [-3.0, 6.0]
plane_from_vec = Plane3D.create_plane_from_vectors_and_point(
vec1, vec2, pt1)
pcd_close = clip_pcd_by_distance_plane(pcd,
plane_from_vec,
threshold,
in_only=True)
pcd_close = pcd_close.extract_low()
seed = 0
np.random.seed(seed)
plane1, _, _, _ = self.sac.ransac(pcd_close.data.T,
constraint=self.plane.param,
constraint_threshold=0.5)
# vis(plane1, pcd, dim_2d=True)
# normal = vec1.reshape([-1,1]) / np.linalg.norm(vec1)
# depth = np.matmul(pcd_close.data.T , normal).reshape([-1])
distance = plane1.distance_to_plane(pcd_close.data.T)
# threshold_outer = 0.3
# threshold_inner = 0.1
# mask_outer = distance < threshold_outer
# mask_inner = distance < threshold_inner
# bin_dist = 5.0
# depth_min = np.min(depth)
# bin_num = int((np.max(depth) - depth_min)/ bin_dist) + 1
# for i in range(bin_num):
# depth_thred_min, depth_thred_max = i*bin_dist+depth_min, (i+1)*bin_dist+depth_min
# mask_depth = np.logical_and(depth > depth_thred_min, depth < depth_thred_max)
# part_inner = np.logical_and(mask_depth, mask_inner)
# part_outer = np.logical_and(mask_depth, mask_outer)
# sum_outer = np.sum(part_outer)
# sum_inner = np.sum(part_inner)
# if sum_outer == 0:
# ratio = 1
# else:
# ratio = float(sum_inner) / sum_outer
# if not ratio == 1:
# print(i, "{:.1f}".format(depth_thred_min), "{:.1f}".format(depth_thred_max), sum_inner, sum_outer, "{:.4f}".format(ratio))
print('Plane params:', plane1.param.T)
threshold_inlier = 0.15
pcd_inlier = pcd_close.data[:, distance <= threshold_inlier]
pcd_outlier = pcd_close.data[:, distance > threshold_inlier]
return plane1, pcd_inlier, pcd_outlier
def create_and_publish_plane_markers(self,
plane,
frame_id='velodyne',
center=None,
normal=None):
if normal is None:
v1 = np.array([[0, 0, 1.0]]).T
else:
v1 = normal
v2 = np.array([[plane.a, plane.b, plane.c]]).T
q_self = Quaternion_self.create_quaternion_from_vector_to_vector(
v1, v2)
q = Quaternion(q_self.x, q_self.y, q_self.z, q_self.w)
euler = np.array(euler_from_quaternion(
(q.x, q.y, q.z, q.w))) * 180 / np.pi
print("Euler: ", euler)
marker_array = MarkerArray()
if center is None:
center = [10, 0, (-plane.a * 10 - plane.d) / plane.c]
marker = visualize_marker(center,
frame_id=frame_id,
mkr_type='cube',
orientation=q,
scale=[20, 10, 0.05],
lifetime=30,
mkr_color=[0.0, 0.8, 0.0, 0.8])
marker_array.markers.append(marker)
return marker_array
# @profile # profiling for analysis
def pcd_callback(self, msg):
rospy.logwarn("Getting pcd at: %d.%09ds, (%d,%d)",
msg.header.stamp.secs, msg.header.stamp.nsecs,
msg.height, msg.width)
pcd_original = pointcloud2_to_xyz_array(msg)
pcd = PointCloud(pcd_original.T)
self.plane, pcd_inlier, pcd_outlier = self.estimate_plane(pcd)
transform_matrix, trans, rot, euler = get_transformation(
frame_from='/velodyne',
frame_to='/world',
time_from=msg.header.stamp,
time_to=msg.header.stamp,
static_frame='/world',
tf_listener=self.tf_listener,
tf_ros=self.tf_ros)
if not transform_matrix is None:
plane_world_param = np.matmul(
np.linalg.inv(transform_matrix).T,
np.array(
[[self.plane.a, self.plane.b, self.plane.c,
self.plane.d]]).T)
plane_world_param = plane_world_param / np.linalg.norm(
plane_world_param[0:3])
if self.plane_tracker is None:
self.plane_tracker = Tracker(msg.header.stamp,
plane_world_param)
else:
self.plane_tracker.predict(msg.header.stamp)
self.plane_tracker.update(plane_world_param)
print("plane_world:", plane_world_param.T)
print("plane_traker:", self.plane_tracker.filter.x_post.T)
# self.plane_world = Plane3D(plane_world_param[0,0], plane_world_param[1,0], plane_world_param[2,0], plane_world_param[3,0])
self.plane_world = Plane3D(self.plane_tracker.filter.x_post[0, 0],
self.plane_tracker.filter.x_post[1, 0],
self.plane_tracker.filter.x_post[2, 0],
self.plane_tracker.filter.x_post[3, 0])
center_pos = np.matmul(
transform_matrix,
np.array([[
10, 0, (-self.plane.a * 10 - self.plane.d) / self.plane.c,
1
]]).T)
center_pos = center_pos[0:3].flatten()
# normal = np.matmul( transform_matrix, np.array([[0., 0., 1., 0.]]).T)
# normal = normal[0:3]
normal = None
marker_array = self.create_and_publish_plane_markers(
self.plane_world,
frame_id='world',
center=center_pos,
normal=normal)
self.pub_plane_markers.publish(marker_array)
plane_msg = Plane()
plane_msg.coef[0], plane_msg.coef[1], plane_msg.coef[
2], plane_msg.coef[
3] = self.plane.a, self.plane.b, self.plane.c, self.plane.d
self.pub_plane.publish(plane_msg)
# pcd_msg_inlier = create_point_cloud(pcd_inlier.T, frame_id='velodyne')
# pcd_msg_outlier = create_point_cloud(pcd_outlier.T, frame_id='velodyne')
pcd_msg_inlier = xyz_array_to_pointcloud2(pcd_inlier.T,
stamp=msg.header.stamp,
frame_id='velodyne')
pcd_msg_outlier = xyz_array_to_pointcloud2(pcd_outlier.T,
stamp=msg.header.stamp,
frame_id='velodyne')
self.pub_pcd_inlier.publish(pcd_msg_inlier)
self.pub_pcd_outlier.publish(pcd_msg_outlier)
rospy.logwarn("Finished plane estimation")
def pose_callback(self, msg):
rospy.logdebug("Getting pose at: %d.%09ds", msg.header.stamp.secs,
msg.header.stamp.nsecs)
centers = []
# y-cooridinate for speed detection line
Y_THRESH = 320
blob_min_width_far = 15
blob_min_height_far = 15
blob_min_width_near = 20
blob_min_height_near = 20
frame_start_time = None
# Create object tracker
tracker = Tracker(80, 3, 2, 1)
# Capture livestream
cap = cv2.VideoCapture(BASE_URL + 'playlist.m3u8')
cap = cv2.VideoCapture('cars.mp4')
while True:
centers = []
frame_start_time = datetime.utcnow()
ret, frame = cap.read()
orig_frame = copy.copy(frame)
# Draw line used for speed detection
cv2.line(frame, (0, Y_THRESH), (1600, Y_THRESH), (255, 0, 0), 2)
data = self.meta_info[b"info"][b"pieces"]
pieces, offset, length = [], 0, len(data)
pieces = [data[offset:offset + 20] for offset in range(0, length, 20)]
return pieces
@property
def output_file(self):
"""
Returns the output filename that we will use to save the torrent data in
"""
logging.info("Torrent output file: {0}".format(
self.meta_info[b"info"][b"name"].decode("utf-8")))
return self.meta_info[b"info"][b"name"].decode("utf-8")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-f",
"--file",
help="Torrent's absolute file path",
type=str,
required=True)
args = parser.parse_args()
t = Torrent(torrent_path=args.file)
tr = Tracker(t)
loop = asyncio.get_event_loop()
r1 = loop.run_until_complete(tr.connect())
print(r1)
def test_board_scan():
t = Tracker(Board(5, 500, 500, 500, othelloAI()))
t.bd.generate_board()
out = t.board_scan()
assert out is False
selection = handle.region()
# Process the first frame
imagefile = handle.frame()
if not imagefile:
sys.exit(0)
# Initialize the Tracker
image = cv2.imread(imagefile, cv2.IMREAD_COLOR)
parameters = Params()
parameters.init_pos = np.floor([selection.x+selection.width/2, selection.y+selection.height/2]) # Initial position
parameters.pos = parameters.init_pos
parameters.target_size = np.floor([selection.width, selection.height])
tracker = Tracker(image, parameters)
tracker.train(image, True)
while True:
# *****************************************
# VOT: Call frame method to get path of the
# current image frame. If the result is
# null, the sequence is over.
# *****************************************
imagefile = handle.frame()
if not imagefile:
break
# Processing the current image
image = cv2.imread(imagefile, cv2.IMREAD_COLOR)
region, lost, xtf = tracker.detect(image)
"-c",
"--config",
action="store",
dest="config",
type="str",
help="Config file, default %default",
default="./config.ini")
options, args = parser.parse_args()
if not os.path.isfile(options.config):
raise RuntimeError('Given config file was not found')
config = Config()
config.read(options.config)
logger = build_logger(config.fetch('logs_dir'), logging.DEBUG)
app = Tracker(config, logger)
app.track()
except KeyboardInterrupt as e:
if logger is not None:
logger.info('Tracker stopped by user. PID: {0}'.format(os.getpid()))
else:
print 'Tracker stopped by user. PID: {0}'.format(os.getpid())
except Exception as e:
if logger is not None:
logger.critical(e.message, exc_info=True)
else:
print traceback.format_exc()
def test_moves_avail():
t = Tracker(Board(5, 500, 500, 500, othelloAI()))
out = t.moves_avail()
assert out is False
class Cloud(object):
"""
Put or get a file from the Cloud. This class maintains the session, which includes all files that we're serving up to other clients.
This class is reponsible for "put"-ting and "get"-ting files to and from the cloud.
"""
def __init__(self, torr_dir="c:/torrent", data_dir="c:/torrent/files", tracker_ip="10.0.0.1:8000", callback=lambda: ".", session_dir="c:/torrent", db_name="torrent.db", ext=".torrent"):
"""
Intialize class.
Parameters
torr_dir: Location for torrent meta data files
data_dir: Location for data files
tracker_ip: IP:PORT string of tracker
callback: option function which will be called while attempting to seed torrents.
torr_db: Name of anydbm database where we save what we're serving.
"""
self.callback = callback
self.data_dir = data_dir
self.torr_dir = torr_dir
self.ext = ext
self.torr_db = os.path.join(session_dir, db_name)
self.db = None # pointer to our torrent DB
self.my_tracker = Tracker(tracker_ip)
self.session = Session(session_dir, db_name)
self.session.register(self.callback)
def put(self, backup_file):
"""
Put files to the tracker, given an encrypted ZIP archive, and then serve them.
Parameters
backup_file: Takes either a full path, or the basename (assumes a directory of self.data_dir in the latter case)
"""
log.debug("backup_file = %s, data_dir = %s, torr_dir = %s, tracker = %s" % (backup_file, self.data_dir, self.torr_dir, self.my_tracker.tracker_ip))
# setup our TorrentMetaInfo object and create a torrent
ti = TorrentMetaInfo(self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, os.path.basename(backup_file))
# Make the torrent file and save new TorrentMetaInfo object with torrent name
if ti.is_valid_create_object():
t=CreateTorrent(ti)
ti = t.create(self.my_tracker)
else:
raise Exception("TorrentMetaInfo object is not valid for CreateTorrent.")
# Upload torrent to tracker. Return false if the upload fails.
if not self.my_tracker.upload_torrent(ti):
log.debug("Failed to upload torrent to tracker.")
return False
# Serve the torrent
return self.session.serve(ti)
def get(self, torr_hash_l):
"""
Pass a torrent info hash, download files for torrent.
Parameters
torr_hash_l: Either a string or a list. Right now, it only takes a single info hash of the torrent to pull down from the tracker. The
data files associated with the torrent are then downloaded from the cloud.
"""
if self.get_torrents(torr_hash_l):
self.get_files(torr_hash_l)
return True
else:
return False
def get_files(self, torr_hash_l):
"""
Download our files, given a torrent's info hash (string) or a list of torrent info hashes (strings). Used by the get() function to pull down data
files from the cloud.
Parameters
torr_hash_l: string, or list of strings
"""
self.session.serve_torrent_by_hash(torr_hash_l, self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, self.ext)
def serve_torrents(self):
"""
Serve up all torrents in our torr directory. Does not take any parameters, simply checks for all torrents
in our torrent meta data directory, and serves them.
"""
self.session.serve_all_torrents(self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, self.ext)
def get_torrents(self, torr_hash_l):
"""
Pull a torrent down from the tracker by ID.
Parameter
torr_hash_l: list or string of info hashes of torrents to download from the tracker.
"""
flag = 0
if hasattr(torr_hash_l, '__iter__'):
for t in torr_hash_l:
torr_loc = self.my_tracker.download_torrent(t, self.torr_dir)
if not torr_loc:
log.debug("Multiple Torr ID's: unable to get torrent location.")
flag = 1
else:
torr_loc = self.my_tracker.download_torrent(torr_hash_l, self.torr_dir)
if not torr_loc:
log.debug("Single ID: unable to get torrent location.")
flag = 1
if flag:
return False
else:
return True
# Mark for removal ?
def get_files_by_torr_name(self, torr_name_l):
"""
Download files given a torrent file name. This function will likely be going away as it's not part of our workflow at the moment.
Parameters
torr_name_l: string or list of torrent info hashes
"""
# setup our TorrentMetaInfo object and create a torrent
if hasattr(torr_name_l, '__iter__'):
for t in torr_name_l:
ti = TorrentMetaInfo(self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, torr_file=t)
self.session.serve(ti)
else:
ti = TorrentMetaInfo(self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, torr_file=torr_name_l)
self.session.serve(ti)
def del_torrent(self, ih):
"""
Given an info hash, delete a torrent locally
"""
try:
self.db = anydbm.open(self.torr_db, 'c')
except:
return False
for info_hash, filename in self.db.iteritems():
if info_hash == ih:
self.unstor_torr(ih)
if self.session.unserve(info_hash=ih):
log.debug("Torrent unserved.")
else:
log.debug("Failed to unserve torrent.")
return True
self.db.close()
return False
###########################################
# Start/Stop Cloud
###########################################
# incomplete
def stop(self):
"""
Save a session's state
"""
# loop through our handles and write out fast resume data
for h in handles:
if h.is_valid() and h.has_metadata():
data = lt.bencode(h.write_resume_data())
open(os.path.join(options.save_path, h.get_torrent_info().name() + '.fastresume'), 'wb').write(data)
# save session settings here
def start(self):
"""
Start the cloud from a saved state
"""
if self.serving_files():
log.debug("Cloud already started.")
return False
# Open our database, and grind through it.
try:
self.db = anydbm.open(self.torr_db, 'c')
except:
return False
flag = 0
for info_hash, filename in self.db.iteritems():
log.debug("Loading... %s %s..." % (info_hash, filename,))
print os.path.join(self.torr_dir, filename+self.ext)
ti = TorrentMetaInfo(self.torr_dir, self.data_dir, self.my_tracker.tracker_ip, filename, str(filename+self.ext))
# if we have it, and the tracker has it, serve it
if not ti.is_valid_torr():
log.debug("Torrent invalid: %s, %s." % (info_hash, filename, ))
self.unstor_torr(info_hash)
elif not self.my_tracker.has_torrent(ti):
log.debug("Torrent does not exist on tracker: %s, %s." % (str(ti.info_hash), filename, ))
self.unstor_torr(info_hash)
else:
self.session.serve(ti)
flag = 1 # serve at least one torrent...
self.db.close()
if flag:
return True
else:
log.debug("Nothing in the database to serve.")
return False
def serving_files(self):
"""
Tells us if the cloud is serving any files.
"""
if len(self.session.handles) > 0:
return True
else:
return False
###########################################
# Database functions
###########################################
def stor_torr(self, info_hash, torr_name):
"""
Store torrents we're serving in the database. If it's already in our database, just
ignore it. Pairs are stored as key => infohash, value=>torr_name
Parameters
torr_name: file name of torrent (string)
info_hash: info hash of torrent (string)
"""
self.db = anydbm.open(self.torr_db, 'c')
if not self.db.has_key(info_hash):
self.db[info_hash] = torr_name
log.debug("info_hash %s, torr_name: %s" % (info_hash, torr_name,))
self.db.close()
def unstor_torr(self, info_hash):
"""
Delete a torrent from our database.
Parameters
info_hash: info hash of torrent to delete
Returns
True if deleted.
False if not deleted, or failure.
"""
try:
self.db = anydbm.open(self.torr_db, 'w')
except:
log.debug("Failed opening database for key removal.")
return False
else:
try:
del self.db[info_hash]
except:
log.debug("Failed removing key.")
return False
else:
return True
def show_db(self):
"""
Enumerate all values in our db
"""
self.db = anydbm.open(self.torr_db, 'c')
for info_hash, filename in self.db.iteritems():
print info_hash, " ", filename
self.db.close()
def setUp(self):
"""Set up a new database session."""
session = create_database_session('sqlite:///:memory:')
self._tracker = Tracker(session)
class TrackerTest(unittest.TestCase):
"""Unit test for the tracker"""
def setUp(self):
"""Set up a new database session."""
session = create_database_session('sqlite:///:memory:')
self._tracker = Tracker(session)
def test_empty_categories(self):
has_category = self._tracker.has_category("anything")
self.assertFalse(has_category)
def test_one_category(self):
self._tracker.create_category('one')
self.assertTrue(self._tracker.has_category("one"))
self.assertFalse(self._tracker.has_category("two"))
def test_two_categories(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
self.assertTrue(self._tracker.has_category("one"))
self.assertTrue(self._tracker.has_category("two"))
def test_duplicated_category_creation(self):
self._tracker.create_category('one')
with self.assertRaises(ValueError):
self._tracker.create_category('one')
def test_default_category_type(self):
self._tracker.create_category('one')
self.assertFalse(self._tracker.is_integer_category('one'))
def test_non_integer_category(self):
self._tracker.create_category('one', is_integer=False)
self.assertFalse(self._tracker.is_integer_category('one'))
def test_empty_categories_listing(self):
self.assertEqual(self._tracker.list_categories(), [])
def test_multiple_categories_listing(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
self.assertEqual(self._tracker.list_categories(), ['one', 'two'])
def test_category_renaming(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
self._tracker.rename_category('two', 'three')
self.assertEqual(self._tracker.list_categories(), ['one', 'three'])
def test_missing_category_renaming(self):
self._tracker.create_category('one')
with self.assertRaises(ValueError):
self._tracker.rename_category('two', 'three')
def test_invalid_category_renaming(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
with self.assertRaises(ValueError):
self._tracker.rename_category('two', 'one')
def test_value_type_correction(self):
self._tracker.create_category('one')
self._tracker.correct_value_type('one', is_integer=True)
self.assertTrue(self._tracker.is_integer_category('one'))
def test_value_correction_to_integer(self):
self._tracker.create_category('one', is_integer=True)
self._tracker.correct_value_type('one', is_integer=False)
self.assertFalse(self._tracker.is_integer_category('one'))
def test_remove_category(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
self._tracker.remove_category('one')
self.assertEqual(self._tracker.list_categories(), ['two'])
def test_remove_invalid_category(self):
self._tracker.create_category('one')
self._tracker.create_category('two')
with self.assertRaises(ValueError):
self._tracker.remove_category('three')
def test_save_measurement(self):
self._tracker.create_category('scores')
self._tracker.save_measurement('scores', 5)
def test_missing_category_for_measurement(self):
with self.assertRaises(ValueError):
self._tracker.save_measurement('missing', 1)
def test_empty_category_listing(self):
self._tracker.create_category('scores')
self.assertEqual(self._tracker.list_measurements('scores'), [])
def test_measurement_listing(self):
self._tracker.create_category('scores')
self._tracker.save_measurement('scores', 5)
self._tracker.save_measurement('scores', 6)
self._tracker.save_measurement('scores', 7)
measurements = self._tracker.list_measurements('scores')
self.assertEqual(measurements[0].id, 1)
self.assertEqual(measurements[1].id, 2)
self.assertEqual(measurements[2].id, 3)
self.assertEqual(measurements[0].value, 5)
self.assertEqual(measurements[1].value, 6)
self.assertEqual(measurements[2].value, 7)
def test_missing_category_measurement_listing(self):
with self.assertRaises(ValueError):
_ = self._tracker.list_measurements('missing')
def test_measurement_correction(self):
self._tracker.create_category('scores')
self._tracker.save_measurement('scores', 5)
self._tracker.save_measurement('scores', 6)
self._tracker.save_measurement('scores', 7)
self._tracker.correct_measurement(1, 1)
self._tracker.correct_measurement(2, 2)
self._tracker.correct_measurement(3, 3)
measurements = self._tracker.list_measurements('scores')
self.assertEqual(measurements[0].id, 1)
self.assertEqual(measurements[1].id, 2)
self.assertEqual(measurements[2].id, 3)
self.assertEqual(measurements[0].value, 1)
self.assertEqual(measurements[1].value, 2)
self.assertEqual(measurements[2].value, 3)
def test_remove_measurement(self):
self._tracker.create_category('scores')
self._tracker.save_measurement('scores', 5)
self._tracker.save_measurement('scores', 6)
self._tracker.save_measurement('scores', 7)
self._tracker.remove_measurement(2)
measurements = self._tracker.list_measurements('scores')
self.assertEqual(measurements[0].id, 1)
self.assertEqual(measurements[1].id, 3)
self.assertEqual(measurements[0].value, 5)
self.assertEqual(measurements[1].value, 7)
def test_remove_missing_measurement(self):
self._tracker.create_category('scores')
with self.assertRaises(ValueError):
self._tracker.remove_measurement(1)
cap = cv.VideoCapture(path)
cap.set(3, WIDTH)
cap.set(4, HEIGHT)
cv.namedWindow('Main')
cv.resizeWindow('Main', (WIDTH, HEIGHT))
total_frames = int(cap.get(cv.CAP_PROP_FRAME_COUNT))
print('ok 1')
initial_lh, initial_hh = 120,220
initial_ls, initial_hs = 122, 140
initial_lv, initial_hv = 135, 160
t = Tracker(initial_lh, initial_hh, initial_ls, initial_hs, initial_lv, initial_hv)
# TRACKER INIT
cv.namedWindow('Tracker')
cv.resizeWindow('Tracker', 600,600)
cv.createTrackbar("L-H", 'Tracker', initial_lh, 255, lambda eventValue, name='l_h': t.update(value=eventValue, name=name))
cv.createTrackbar("H-H", 'Tracker', initial_hh, 360, lambda eventValue, name='h_h': t.update(value=eventValue, name=name))
cv.createTrackbar("L-S", 'Tracker', initial_ls, 255, lambda eventValue, name='l_s': t.update(value=eventValue, name=name))
cv.createTrackbar("H-S", 'Tracker', initial_hs, 255, lambda eventValue, name='h_s': t.update(value=eventValue, name=name))
cv.createTrackbar("L-V", 'Tracker', initial_lv, 255, lambda eventValue, name='l_v': t.update(value=eventValue, name=name))
cv.createTrackbar("H-V", 'Tracker', initial_hv, 255, lambda eventValue, name='h_v': t.update(value=eventValue, name=name))
cv.createTrackbar("Thereshold_Min", 'Tracker', 127, 255, lambda eventValue, name='threshold_min': t.update(value=eventValue, name=name))
cv.createTrackbar("Thereshold_Max", 'Tracker', 255, 255, lambda eventValue, name='threshold_man': t.update(value=eventValue, name=name))
class SVMCLF():
def __init__(self):
# ROS nodes
rospy.init_node('classifier', anonymous=True)
rospy.Subscriber("camera/image", Image, self.callback, queue_size=1)
self.pub = rospy.Publisher('driver_node/drivestate',
Int8,
queue_size=1,
latch=True)
self.imgcnn_pub = rospy.Publisher('camera/imgcnn', Image, queue_size=1)
# publish "stop" on the drivestate topic
print 'publishing -1'
self.pub.publish(-1)
print 'Importing tensorflow ...'
import tensorflow as tf
print 'Tensorflow imported.'
# publish "go" on the drivestate topic
# Trained SVM model
fn_model = rospy.get_param('~cnnModelFile')
graph_file = rospy.get_param('~cnnGraphFile')
saver = tf.train.import_meta_graph(graph_file)
self.sess = tf.Session()
graph_dir = os.path.dirname(graph_file)
graph = tf.get_default_graph()
saver.restore(self.sess, tf.train.latest_checkpoint(graph_dir))
self.prediction_op = graph.get_tensor_by_name("prediction_op:0")
self.logits_op = graph.get_tensor_by_name("logits_op:0")
#self.probs = graph.get_tensor_by_name("probs:0")
self.x_placeholder = graph.get_tensor_by_name("x:0")
self.keep_prob = graph.get_tensor_by_name("keep_prob:0")
# svmparams = pickle.load(open(fn_params, 'rb')) #pickle.load(f2)
# self.fmean = svmparams['fmean']
# self.fstd = svmparams['fstd']
# Settings
self.nhistory = 1 # tracker buffer
self.dt = 0.1 # update interal
self.K_detthresh_stop = 0.7 # detection threshold (factor of window count)
self.K_detthresh_warn = 1.2 # detection threshold (factor of window count)
self.K_mapthresh = 0.08 # discard threshold (factor of window count)
self.K_stopbias = 0.4 # bias to favor STOP over WARN (factor of window count)
# Updates
self.drive_state = 0
self.img = [None]
self.lastimgtime = -1
# tracker
self.tracker = Tracker(self.nhistory)
# start
self.loop()
def callback(self, rosimg):
self.img = CvBridge().imgmsg_to_cv2(rosimg)
self.lastimgtime = rosimg.header.stamp.secs
def loop(self):
rate = rospy.Rate(1 / self.dt)
lastupdate = -1
while not rospy.is_shutdown():
if (self.lastimgtime != lastupdate):
start_time = time.time()
# ---- process frame ---
rosimg = self.img # copy to local memory before processing
dec, draw_img = self.processOneFrame(rosimg)
lastupdate = self.lastimgtime
# Publish results
#print "State=" + str(dec) + " (" + str(time.time() - start_time) + "s)"
self.drive_state = dec
#print ("Drivestate = ",dec)
self.pub.publish(self.drive_state)
# Optional - Publish image for monitoring
self.imgcnn_pub.publish(CvBridge().cv2_to_imgmsg(
draw_img, "bgr8"))
rate.sleep()
#def getFeatures(self,img):
# return [
# imagefunctions.num_corners(img),
# imagefunctions.num_edges(img),
# imagefunctions.num_red_pixels(img),
# imagefunctions.num_white_pixels(img),
# imagefunctions.abs_sobel_thresh(img, orient='y', sobel_kernel=3, thresh=(100, 200)),
# imagefunctions.mag_thresh(img, sobel_kernel=5, mag_thresh=(100, 180)),
# imagefunctions.dir_threshold(img, sobel_kernel=3, thresh=(np.pi/8, np.pi/4))
# ]
#def normalize_features(self,feature_vector,fmn,fsd):
# numDim = len(feature_vector)
# normFeatures = []
# normfeat = [None]*numDim
# for i in range(numDim):
# normfeat[i] = (feature_vector[i]-fmn[i])/fsd[i]
# normFeatures.append(normfeat)
# #transpose result
# res = np.array(normFeatures).T
# return res
def draw_labeled_bboxes(self, img, labels, boxcolor):
# Iterate through all detected cars
for item_number in range(1, labels[1] + 1):
# Find pixels with each item_number label value
nonzero = (labels[0] == item_number).nonzero()
# Identify x and y values of those pixels
nonzeroy = np.array(nonzero[0])
nonzerox = np.array(nonzero[1])
bbox = ((np.min(nonzerox), np.min(nonzeroy)), (np.max(nonzerox),
np.max(nonzeroy)))
# Draw the box on the image
cv2.rectangle(img, bbox[0], bbox[1], boxcolor, 2)
# Return the image
return img
def search_windows(self, img, windows, framenum=0):
# preprocess frame
img_prep = imagefunctions.preprocess_one_rgb(img[0:127][:])
fvec = []
for window in windows:
# extract test window from image
test_img = img_prep[window[0][1]:window[1][1],
window[0][0]:window[1][0]]
# extract features
#feat = self.getFeatures(test_img)
# normalize features
#normfeat = self.normalize_features(feat,self.fmean,self.fstd)
# assemble batch
#testvec = np.asarray(normfeat).reshape(1,-1)
fvec.append(test_img)
# batch prediction
if (np.array(fvec).ndim >= 3):
#prob_vec = self.sess.run(self.probs, feed_dict={self.x_placeholder:np.asarray(fvec, dtype=np.float32), self.keep_prob:1.0})
rvec = self.sess.run(self.prediction_op,
feed_dict={
self.x_placeholder:
np.asarray(fvec, dtype=np.float32),
self.keep_prob:
1.0
}) #np.array(fvec).squeeze(axis=1))
else:
rvec = []
# list of positive stop sign detection windows
stop_indices = [i for i, x in enumerate(rvec) if x == 1]
stop_windows = [windows[i] for i in stop_indices]
# list of positive warn sign detection windows
warn_indices = [i for i, x in enumerate(rvec) if x == 2]
warn_windows = [windows[i] for i in warn_indices]
# return positve detection windows
return stop_windows, warn_windows
def find_signs(self, img):
startx = 10
stopx = 68 #imgsize[0]-windowsize[0] #80
starty = 0 #20 #19
stopy = imgsize[1] - windowsize[1] #30
window_list = []
for x in range(startx, stopx, slidestep[0]):
for y in range(starty, stopy, slidestep[1]):
img_in = img[y:y + windowsize[1], x:x + windowsize[0]]
#img_crop_pp = imagefunctions.preprocess_one_rgb(img_crop)
#img_in = np.array(255*img_crop_pp, dtype=np.uint8)
if (imagefunctions.num_red_pixels(img_in) > min_red_pixels):
window_list.append(
((x, y), (x + windowsize[0], y + windowsize[1])))
#stop_windows, warn_windows = self.search_windows(img, window_list, framenum=random.randint(0,9999))
stop_windows, warn_windows = self.search_windows(img, window_list)
# if no window to search
numwin = len(window_list)
if (numwin == 0):
decision = 0
labels = [None]
return decision, labels, img
# Method 1 - Count windows
# if ((len(stop_windows)<2) and (len(warn_windows)<2)):
# return 0,[None]
# elif (len(stop_windows)>=len(warn_windows)):
# return 1,[None]
# else:
# return 2,[None]
# Method 2 - Localized heatmap based decision
heat_stop = np.zeros_like(img[:, :, 0]).astype(np.float)
heat_warn = np.zeros_like(img[:, :, 0]).astype(np.float)
for bbox in window_list:
startx = bbox[0][0]
starty = bbox[0][1]
endx = bbox[1][0]
endy = bbox[1][1]
cv2.rectangle(img, (startx, starty), (endx, endy), (200, 0, 0), 1)
for bbox in warn_windows:
startx = bbox[0][0]
starty = bbox[0][1]
endx = bbox[1][0]
endy = bbox[1][1]
heat_warn[starty:endy, startx:endx] += 1.
cv2.rectangle(img, (startx, starty), (endx, endy), (0, 255, 0), 1)
for bbox in stop_windows:
startx = bbox[0][0]
starty = bbox[0][1]
endx = bbox[1][0]
endy = bbox[1][1]
heat_stop[starty:endy, startx:endx] += 1.
cv2.rectangle(img, (startx, starty), (endx, endy), (0, 0, 255), 1)
score_stop = np.max(heat_stop)
score_warn = np.max(heat_warn)
print '[scores] stop:' + str(score_stop) + ' warn:' + str(score_warn)
# ---- GET DECISION ---- #
decision = self.get_decision(score_stop, score_warn, numwin)
# plot final decision region
mapthresh = self.K_mapthresh * numwin
labels = [None]
if (decision == 1):
heatmap_stop = heat_stop
heatmap_stop[heatmap_stop <= mapthresh] = 0
labels = label(heatmap_stop)
elif (decision == 2):
heatmap_warn = heat_warn
heatmap_warn[heatmap_warn <= mapthresh] = 0
labels = label(heatmap_warn)
return decision, labels, img
def get_decision(self, score_stop, score_warn, numwin):
# decision thresholds and biases
detthresh_stop = self.K_detthresh_stop * numwin
detthresh_warn = self.K_detthresh_warn * numwin
#stopbias = self.K_stopbias * numwin
print 'numwin = ' + str(numwin) + ', detthreshSTOP = ' + str(
detthresh_stop) + ', detthreshWARN = ' + str(detthresh_warn)
# Make Decision
if (score_stop >=
detthresh_stop): # and (score_stop + stopbias > score_warn):
decision = 1
elif (score_warn >= detthresh_warn):
decision = 2
else:
decision = 0
return decision
def processOneFrame(self, img):
# get decision for current frame
dec, labels, draw_img = self.find_signs(img)
# combine with previous results (if applicable)
self.tracker.new_data(dec)
final_decision = self.tracker.combined_results()
# return results and output image
return final_decision, draw_img
def main():
# Create opencv video capture object
cap = cv2.VideoCapture('G:/cmu/colonoscopy/New folder/Cold.mp4')
#cap = cv2.VideoCapture('G:/cmu/colonoscopy/imagemark/Color-Tracker-master/Retroflect-at-end.mp4')
# Create Object Detector
detector = Detectors()
# Create Object Tracker
tracker = Tracker(160, 1000, 5, 100)
# Variables initialization
skip_frame_count = 0
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
pause = False
num = 0
frame_num = 0
# Infinite loop to process video frames
while (True):
frame_num += 1
print(frame_num)
# Capture frame-by-frame
ret, frame = cap.read()
frame = frame[30:550, 400:930]
#frame = frame[40:400,130:450]
# Make copy of original frame
orig_frame = copy.copy(frame)
# Skip initial frames that display logo
if (skip_frame_count < 15):
skip_frame_count += 1
continue
# Detect and return centeroids of the objects in the frame
centers = detector.Detect1(orig_frame)
# If centroids are detected then track them
if (len(centers) > 0):
text = 'Biopsy'
cv2.putText(orig_frame,
text, (100, 100),
cv2.FONT_HERSHEY_SIMPLEX,
1.0, (0, 0, 255),
2,
lineType=cv2.LINE_AA)
# Track object using Kalman Filter
tracker.Update(centers)
# For identified object tracks draw tracking line
# Use various colors to indicate different track_id
for i in range(len(tracker.tracks)):
if (len(tracker.tracks[i].trace) > 1):
for j in range(len(tracker.tracks[i].trace) - 1):
# Draw trace line
x1 = tracker.tracks[i].trace[j][0][0]
y1 = tracker.tracks[i].trace[j][1][0]
x2 = tracker.tracks[i].trace[j + 1][0][0]
y2 = tracker.tracks[i].trace[j + 1][1][0]
clr = tracker.tracks[i].track_id % 9
cv2.line(frame, (int(x1), int(y1)), (int(x2), int(y2)),
track_colors[clr], 2)
# Display the resulting tracking frame
cv2.imshow('Tracking', frame)
# Display the original frame
cv2.imshow('Original', orig_frame)
print(num)
# Slower the FPS
cv2.waitKey(20)
# Check for key strokes
k = cv2.waitKey(50) & 0xff
if k == 27: # 'esc' key has been pressed, exit program.
break
if k == 112: # 'p' has been pressed. this will pause/resume the code.
pause = not pause
if (pause is True):
print("Code is paused. Press 'p' to resume..")
while (pause is True):
# stay in this loop until
key = cv2.waitKey(30) & 0xff
if key == 112:
pause = False
print("Resume code..!!")
break
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
def main():
global args, device
parser = argparse.ArgumentParser(
description='Tracking Video Visualization')
parser.add_argument('--video_path',
default='test_data/ohaqlzfnuv.mp4',
help='video path')
parser.add_argument('--ground_truth', help='ground truth file path')
parser.add_argument('--model',
default='models/SiamRPNOTB.model',
help='model path')
parser.add_argument('--net_name',
type=str,
default='SiamRPNotb',
help='network, SiamRPNbatchOTBMobile, SiamRPNotb')
parser.add_argument('--interval',
type=int,
default=-1,
help='initialize interval')
parser.add_argument('--mode',
type=str,
default='test',
help='[test, eval]')
parser.add_argument('--result_video', help='result video')
parser.add_argument('--result_file',
default='saved_data/result.txt',
help='result file')
parser.add_argument('-v',
'--visualization',
dest='visualization',
action='store_true',
help='whether visualize result')
args = parser.parse_args()
args.visualization = False
tracker = Tracker(0, args.net_name, args.model)
if args.mode == 'eval' and args.ground_truth is not None:
with open(args.ground_truth) as f:
labels = f.readlines()
gt_rects = []
for label in labels:
label = re.split(';|,|\t| |, |; ', label.strip('\n'))
label = list(map(int, label))
gt_rects.append(label)
# Todo: multi video test
thresholds_overlap = np.arange(0, 1.05, 0.05)
success_overlap = np.zeros(len(thresholds_overlap))
fps, regions = eval_video(gt_rects, args.interval, tracker)
gt_rects = get_none_rotation_rect(np.array(gt_rects))
gt_rects = corner_to_rect(gt_rects)
success_overlap = compute_success_overlap(gt_rects, np.array(regions))
auc = success_overlap.mean()
logging.info(
'Mean Running AP:{:.4f} Mean Running Speed {:.1f}fps'.format(
auc, fps))
elif args.mode == 'test':
assert (args.interval == -1)
fps, regions = test_video(tracker)
print(fps)
# object detection
img_size = (tracktor['width'], tracktor['height'])
backbone = resnet_fpn_backbone(tracktor['backbone'], True)
backbone.out_channels = 256
obj_detect = Jde_RCNN(backbone, num_ID=tracktor['num_ID'], min_size=img_size[1], max_size=img_size[0], version=tracktor['version'])
checkpoint = torch.load(tracktor['weights'], map_location='cpu')['model']
# if tracktor['version']=='v2':
# checkpoint['roi_heads.embed_extractor.extract_embedding.weight'] = checkpoint['roi_heads.box_predictor.extract_embedding.weight']
# checkpoint['roi_heads.embed_extractor.extract_embedding.bias'] = checkpoint['roi_heads.box_predictor.extract_embedding.bias']
print(obj_detect.load_state_dict(checkpoint, strict=False))
obj_detect.eval()
obj_detect.cuda()
tracker = Tracker(obj_detect, tracktor['tracker'])
transforms = T.Compose([T.ToTensor()])
time_total = 0
num_frames = 0
mot_accums = []
for seq_path in os.listdir(tracktor['dataset']):
tracker.reset()
start = time.time()
print(f"Tracking: {seq_path}")
sequence = LoadImagesAndLabels(root, osp.join(tracktor['dataset'], seq_path), img_size, augment=False, transforms=transforms)
data_loader = DataLoader(sequence, batch_size=1, shuffle=False)
class Phaser(pyglet.window.Window):
MODE_NORMAL = 0
MODE_TRACING = 1
MODE_CREATING = 2
def __init__(self):
super(Phaser, self).__init__(
WIDTH,
HEIGHT) # config=pyglet.gl.Config(sample_buffers=1, samples=2))
self.xdim = WIDTH
self.ydim = HEIGHT
self.load_fonts()
self.gesture_sound = pyglet.media.StaticSource(
pyglet.resource.media('gesture.wav'))
# add the faders
self.faders = []
self.title_fader = Fader(0.1, 0.9)
self.faders.append(self.title_fader)
self.mouse = (0, 0)
self.title_image = pyglet.font.Text(self.fonts[40], 'PhaseSeq')
self.mode_image = pyglet.font.Text(self.fonts[30], 'Trace mode')
self.show_saved_image = pyglet.font.Text(self.fonts[20],
'Showing saved traces')
self.create_image = pyglet.font.Text(self.fonts[30],
'Creating classifier')
self.showinst_image = pyglet.font.Text(
self.fonts[20], 'Press H to toggle instructions')
instructions = 'To create a classifier:\n'
instructions += '- Right click to begin\n'
instructions += '- Left click to place first path segment\n'
instructions += '- Left click again to extend path\n'
instructions += '- Optionally use <A> and <D> to change width\n'
instructions += '- Repeat until path complete\n'
instructions += '- Right click to save\n'
instructions += '(Press <U> to undo last point while drawing)\n\n'
instructions += 'To preserve traces on screen:\n'
instructions += '- Hold <Shift> and perform a movement\n'
instructions += '- Release <Shift> to stop recording\n'
instructions += '- Perform these steps as many times as needed\n'
instructions += '- Press <T> to toggle showing all stored recordings\n'
instructions += '- Press <X> to clear all stored recordings\n'
instructions += '\nOther keybindings:\n'
instructions += '<C> clears all saved classifiers\n'
instructions += '<L> loads saved classifiers from file\n'
instructions += '<S> saves current set of classifiers to file\n'
self.inst_label = pyglet.text.Label(instructions,
font_size=16,
x=self.width / 2,
y=self.height - 100,
anchor_x='center',
multiline=True,
width=self.width * 0.75)
self.phase_trace = []
self.trace_len = 150
self.sequences = []
self.active_sequence = None
self.saved_traces = []
self.show_saved_traces = False
self.extents = [-0.05, 1.1, -0.6, 0.6]
self.thickness = 0.07
self.tracker = Tracker(WIDTH, HEIGHT)
self.mode = Phaser.MODE_NORMAL
self.show_instructions = False
self.import_sequences()
def start_sequence(self):
self.active_sequence = []
def end_sequence(self):
if len(self.active_sequence) > 1:
self.sequences.append(
TriPhaseSequence(
self.make_path(self.active_sequence, self.thickness)))
self.active_sequence = None
def import_sequences(self):
if not os.path.exists(SEQUENCES_FILE):
print('No existing sequences file found!')
return
self.sequences = []
self.active_sequence = None
with open(SEQUENCES_FILE, 'r') as f:
seq_data = f.readlines()
for sd in seq_data:
points = []
sd = sd.split(',')
for i in range(0, len(sd), 6):
t = map(float, sd[i:i + 6])
points.append(((t[0], t[1]), (t[2], t[3]), (t[4], t[5])))
self.sequences.append(TriPhaseSequence(points))
print('Imported %d sequences' % (len(self.sequences)))
def export_sequences(self):
if len(self.sequences) == 0:
print('No sequences to export')
return
print('Exporting %d sequences to %s' %
(len(self.sequences), SEQUENCES_FILE))
with open(SEQUENCES_FILE, 'w') as f:
for j, s in enumerate(self.sequences):
points = list(itertools.chain.from_iterable(s.states))
for i, p in enumerate(points):
f.write('%f,%f' % p)
if i < len(points) - 1:
f.write(',')
else:
if j < len(self.sequences) - 1:
f.write('\n')
print('Sequence exported (%d points)' % (len(points) / 3))
def load_fonts(self):
# load fonts
self.fonts = {}
pyglet.font.add_file("rez.ttf")
for i in range(6, 41):
self.fonts[i] = pyglet.font.load('Rez', i)
def start(self):
pyglet.clock.schedule_interval(self.update, 1 / 60.0)
pyglet.app.run()
self.export_sequences()
def update(self, dt):
"""Frame update"""
# update the faders, so they fade in and out
for fader in self.faders:
fader.update(dt)
self.tracker.update(pos=self.mouse)
n = len(self.tracker.the_object.d_seq)
for d in self.tracker.the_object.d_seq:
state = (d[0, 0], -d[0, 1] * 4)
self.phase_trace.append(state)
if self.mode == Phaser.MODE_NORMAL:
for i, seq in enumerate(self.sequences):
seq.update(state[0], state[1], dt / n)
if seq.active == 1.0:
self.gesture_sound.play()
self.title_image = pyglet.font.Text(
self.fonts[40], 'Gesture %d' % i)
self.title_fader.reset()
if self.mode != Phaser.MODE_TRACING and len(
self.phase_trace) > self.trace_len:
self.phase_trace = self.phase_trace[len(self.phase_trace) -
self.trace_len:]
self.draw()
def draw_axes(self):
w = 0.8 * self.extents[1]
y1 = 0.7 * self.extents[2]
y2 = 0.7 * self.extents[3]
glLineWidth(3.0)
glEnable(GL_BLEND)
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
if self.mode == Phaser.MODE_TRACING:
glColor4f(0.2, 0.9, 0.5, 0.3)
else:
glColor4f(0.2, 0.5, 0.9, 0.3)
glBegin(GL_LINES)
glVertex2f(0, y1)
glVertex2f(0, y2)
glVertex2f(0, 0)
glVertex2f(w, 0)
glEnd()
if self.mode == Phaser.MODE_TRACING:
glColor4f(0.2, 0.9, 0.6, 0.1)
glBegin(GL_QUADS)
glVertex2f(0, 0)
glColor4f(0.2, 0.9, 0.6, 0.0)
glVertex2f(0, y1)
glVertex2f(w, y1)
glColor4f(0.2, 0.9, 0.6, 0.1)
glVertex2f(w, 0)
else:
glColor4f(0.2, 0.6, 0.9, 0.1)
glBegin(GL_QUADS)
glVertex2f(0, 0)
glColor4f(0.2, 0.6, 0.9, 0.0)
glVertex2f(0, y1)
glVertex2f(w, y1)
glColor4f(0.2, 0.6, 0.9, 0.1)
glVertex2f(w, 0)
glEnd()
if self.mode == Phaser.MODE_TRACING:
glColor4f(0.2, 0.9, 0.5, 0.4)
else:
glColor4f(0.2, 0.5, 0.9, 0.4)
n = 10
for i in range(n):
x = w * ((i + 1) / float(n))
glBegin(GL_LINES)
glVertex2f(x, -0.01)
glVertex2f(x, 0.01)
glEnd()
glLineWidth(1.0)
def draw_trace(self):
glBlendFunc(GL_SRC_ALPHA, GL_ONE)
k = 1.0
glLineWidth(3.0)
glBegin(GL_LINE_STRIP)
for vertex in reversed(self.phase_trace):
if vertex:
glColor4f(0.9, 0.7, 0.7, k)
glVertex2f(vertex[0], vertex[1])
if self.mode != Phaser.MODE_TRACING:
k = k * 0.95 #0.9
glEnd()
glLineWidth(1.0)
def draw_tri(self, tri, color, phase=0):
alpha = color[3]
glColor4f(color[0], color[1], color[2], alpha) #/2.0)
glBegin(GL_TRIANGLES)
glVertex2f(*tri[0])
glVertex2f(*tri[1])
glVertex2f(*tri[2])
glEnd()
# glBegin(GL_LINE_LOOP)
# glColor4f(color[0], color[1], color[2], alpha)
# glVertex2f(*tri[0])
# glVertex2f(*tri[1])
# glVertex2f(*tri[2])
# glEnd()
def draw_saved_traces(self):
if not self.show_saved_traces:
return
glBlendFunc(GL_SRC_ALPHA, GL_ONE)
glLineWidth(3.0)
colours = [[0.9, 0.2, 0.2, 0.8], [0.2, 0.9, 0.2, 0.8],
[0.2, 0.2, 0.9, 0.8]]
for i, st in enumerate(self.saved_traces):
glBegin(GL_LINE_STRIP)
glColor4f(*colours[i % 3])
for j, vertex in enumerate(st):
glVertex2f(vertex[0], vertex[1])
glEnd()
glLineWidth(1.0)
def draw_sequences(self):
for sequence in self.sequences:
for i, elt in enumerate(sequence.states):
if sequence.active > 0.1:
self.draw_tri(elt, (1, 1, 1, sequence.active))
else:
if i == sequence.state:
self.draw_tri(elt,
(0.6, 0.9, 0.6, sequence.trigger + 0.1))
else:
self.draw_tri(elt, (0.6, 0.9, 0.6, 0.25), phase=i)
if self.active_sequence:
self.draw_thick_line(self.active_sequence)
def make_path(self, seq, thickness):
"""Convert a linear sequence to a set of triangles"""
last = None
last_l = None
tris = []
#if len(seq)>2:
# spl = spline.CardinalSpline(seq, tension=0.2)
# seq = [spl(x/3.0) for x in range(3*(len(seq)-1))]
for pt in seq:
if last:
normal = geometry.normal(last, pt)
strut = geometry.mul(normal, thickness)
if not last_l:
last_r = (geometry.sub(strut, last))
last_l = (geometry.add(strut, last))
tris.append((geometry.add(strut, pt), last_r, last_l))
tris.append(
(geometry.add(strut, pt), geometry.sub(strut, pt), last_r))
last_l = geometry.add(strut, pt)
last_r = geometry.sub(strut, pt)
last = pt
return tris
def draw_thick_line(self, seq):
glColor4f(0.3, 0.8, 0.2, 0.5)
if len(seq) == 1:
glPointSize(4.0)
glBegin(GL_POINTS)
glVertex2f(*seq[0])
glEnd()
return
tris = self.make_path(seq, self.thickness)
for tri in tris:
glBegin(GL_LINE_LOOP)
for vertex in tri:
glVertex2f(*vertex)
glEnd()
def draw_tracing_background(self):
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
glBegin(GL_QUADS)
glColor4f(0.33, 0.85, 0.26, 0.3)
glVertex2f(0, 0)
glColor4f(0.33, 0.85, 0.26, 0.3)
glVertex2f(self.width, 0)
glColor4f(0.55, 0.85, 0.46, 0.3)
glVertex2f(self.width, self.height)
glColor4f(0.55, 0.85, 0.46, 0.3)
glVertex2f(0, self.height)
glEnd()
def draw_background(self):
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
glBegin(GL_QUADS)
glColor4f(0.02, 0.04, 0.12, 1)
glVertex2f(0, 0)
glColor4f(0.02, 0.04, 0.12, 1)
glVertex2f(self.width, 0)
glColor4f(0.03, 0.07, 0.22, 1)
glVertex2f(self.width, self.height)
glColor4f(0.13, 0.07, 0.22, 1)
glVertex2f(0, self.height)
glEnd()
def draw(self):
"""Draw the entire screen"""
glClearColor(0.02, 0.04, 0.12, 1)
self.last_frame_time = time.clock()
self.clear()
if self.mode == Phaser.MODE_TRACING:
self.draw_tracing_background()
else:
self.draw_background()
glLoadIdentity()
glMatrixMode(GL_PROJECTION)
glPushMatrix()
glLoadIdentity()
glOrtho(self.extents[0], self.extents[1], self.extents[2],
self.extents[3], -1, 1)
glMatrixMode(GL_MODELVIEW)
glLoadIdentity()
self.draw_axes()
self.draw_trace()
self.draw_saved_traces()
self.draw_sequences()
glMatrixMode(GL_PROJECTION)
glPopMatrix()
glMatrixMode(GL_MODELVIEW)
glLoadIdentity()
self.draw_title()
self.draw_mode()
self.draw_instructions()
def on_key_press(self, symbol, modifiers):
if self.mode == Phaser.MODE_NORMAL and symbol == pyglet.window.key.LSHIFT or symbol == pyglet.window.key.RSHIFT:
self.mode = Phaser.MODE_TRACING
self.phase_trace = []
pyglet.window.Window.on_key_press(self, symbol, modifiers)
def on_key_release(self, symbol, modifiers):
if symbol == pyglet.window.key.C:
print('Clearing sequences')
self.sequences = []
self.active_sequence = None
if os.path.exists(SEQUENCES_FILE):
os.unlink(SEQUENCES_FILE)
elif symbol == pyglet.window.key.A:
self.thickness -= 0.01
print('thickness =', self.thickness)
elif symbol == pyglet.window.key.D:
self.thickness += 0.01
print('thickness =', self.thickness)
elif self.mode == Phaser.MODE_TRACING and symbol == pyglet.window.key.LSHIFT or symbol == pyglet.window.key.RSHIFT:
self.mode = Phaser.MODE_NORMAL
self.saved_traces.append(self.phase_trace)
self.phase_trace = []
elif symbol == pyglet.window.key.S:
self.export_sequences()
elif symbol == pyglet.window.key.L:
self.import_sequences()
elif symbol == pyglet.window.key.H:
self.show_instructions = not self.show_instructions
elif symbol == pyglet.window.key.U:
if self.mode == Phaser.MODE_CREATING:
if len(self.active_sequence) > 1:
self.active_sequence = self.active_sequence[:-1]
else:
self.end_sequence()
self.mode = Phaser.MODE_NORMAL
elif symbol == pyglet.window.key.X:
self.saved_traces = []
print('cleared saved traces')
elif symbol == pyglet.window.key.T:
self.show_saved_traces = not self.show_saved_traces
def on_mouse_motion(self, x, y, dx, dy):
self.mouse = (x, y)
def new_state(self, x, y):
x = x / float(self.width)
y = y / float(self.height)
rx = (1 - x) * self.extents[0] + (x) * self.extents[1]
ry = (1 - y) * self.extents[2] + (y) * self.extents[3]
self.active_sequence.append((rx, ry))
def on_mouse_release(self, x, y, button, modifiers):
self.mouse = (x, y)
# left mouse click adds new segment to current sequence
if button == pyglet.window.mouse.LEFT and self.active_sequence != None:
self.new_state(x, y)
return
# right mouse ends an active sequence...
if (button == pyglet.window.mouse.RIGHT \
or (button == pyglet.window.mouse.LEFT and modifiers & 2)) \
and self.active_sequence != None:
self.end_sequence()
self.mode = Phaser.MODE_NORMAL
return
# ...or starts a new one
if (button == pyglet.window.mouse.RIGHT \
or (button == pyglet.window.mouse.LEFT and modifiers & 2)) \
and self.active_sequence == None:
self.start_sequence()
self.mode = Phaser.MODE_CREATING
self.new_state(x, y)
return
def draw_instructions(self):
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
self.showinst_image.x = 10
self.showinst_image.y = 30
self.showinst_image.halign = 'left'
self.showinst_image.valign = 'center'
self.showinst_image.color = (1, 1, 1, 1)
self.showinst_image.draw()
if self.show_instructions:
self.inst_label.draw()
def draw_mode(self):
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
if self.mode == Phaser.MODE_TRACING:
self.mode_image.x = self.width / 2
self.mode_image.y = self.height - 40
self.mode_image.halign = "center"
self.mode_image.valign = "center"
self.mode_image.color = (1, 1, 1, 1)
self.mode_image.draw()
elif self.mode == Phaser.MODE_CREATING:
self.create_image.x = self.width / 2
self.create_image.y = self.height - 40
self.create_image.halign = "center"
self.create_image.valign = "center"
self.create_image.color = (1, 1, 1, 1)
self.create_image.draw()
if self.show_saved_traces:
self.show_saved_image.x = self.width - 10
self.show_saved_image.y = 30
self.show_saved_image.halign = 'right'
self.show_saved_image.valign = 'center'
self.show_saved_image.color = (1, 1, 1, 1)
self.show_saved_image.draw()
def draw_title(self):
"""Draw the title, fading in and out"""
k = self.title_fader.get()
if k > 1e-3:
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA)
self.title_image.x = self.width / 2
self.title_image.y = self.height / 2
self.title_image.halign = "center"
self.title_image.valign = "center"
self.title_image.color = (1, 1, 1, k)
self.title_image.draw()
input_dim = (384, 512, 3)
K.set_learning_phase(0) # 0 testing, 1 training mode
field_width = 151
field_height = 91
radius = 5.5
blob_min_width = 4
blob_min_height = 4
boundary_thresh = 75
frame_start_time = None
# Create object tracker
tracker = Tracker(200, 4, 3, 1)
hflip = 1
vflip = 0
if hflip and vflip:
c = -1
else:
c = 0 if vflip else 1
# Capture livestream
cap = VideoCapture("http://192.168.43.1:8080/video")
# cap = VideoCapture(0)
ds = DataSender("127.0.0.1", 1835)
dr = DataReciever("127.0.0.1", 1836)
def __init__(self):
super(Phaser, self).__init__(
WIDTH,
HEIGHT) # config=pyglet.gl.Config(sample_buffers=1, samples=2))
self.xdim = WIDTH
self.ydim = HEIGHT
self.load_fonts()
self.gesture_sound = pyglet.media.StaticSource(
pyglet.resource.media('gesture.wav'))
# add the faders
self.faders = []
self.title_fader = Fader(0.1, 0.9)
self.faders.append(self.title_fader)
self.mouse = (0, 0)
self.title_image = pyglet.font.Text(self.fonts[40], 'PhaseSeq')
self.mode_image = pyglet.font.Text(self.fonts[30], 'Trace mode')
self.show_saved_image = pyglet.font.Text(self.fonts[20],
'Showing saved traces')
self.create_image = pyglet.font.Text(self.fonts[30],
'Creating classifier')
self.showinst_image = pyglet.font.Text(
self.fonts[20], 'Press H to toggle instructions')
instructions = 'To create a classifier:\n'
instructions += '- Right click to begin\n'
instructions += '- Left click to place first path segment\n'
instructions += '- Left click again to extend path\n'
instructions += '- Optionally use <A> and <D> to change width\n'
instructions += '- Repeat until path complete\n'
instructions += '- Right click to save\n'
instructions += '(Press <U> to undo last point while drawing)\n\n'
instructions += 'To preserve traces on screen:\n'
instructions += '- Hold <Shift> and perform a movement\n'
instructions += '- Release <Shift> to stop recording\n'
instructions += '- Perform these steps as many times as needed\n'
instructions += '- Press <T> to toggle showing all stored recordings\n'
instructions += '- Press <X> to clear all stored recordings\n'
instructions += '\nOther keybindings:\n'
instructions += '<C> clears all saved classifiers\n'
instructions += '<L> loads saved classifiers from file\n'
instructions += '<S> saves current set of classifiers to file\n'
self.inst_label = pyglet.text.Label(instructions,
font_size=16,
x=self.width / 2,
y=self.height - 100,
anchor_x='center',
multiline=True,
width=self.width * 0.75)
self.phase_trace = []
self.trace_len = 150
self.sequences = []
self.active_sequence = None
self.saved_traces = []
self.show_saved_traces = False
self.extents = [-0.05, 1.1, -0.6, 0.6]
self.thickness = 0.07
self.tracker = Tracker(WIDTH, HEIGHT)
self.mode = Phaser.MODE_NORMAL
self.show_instructions = False
self.import_sequences()
class TrackerClient(discord.Client):
def __init__(self, *args, **kwargs):
super(TrackerClient, self).__init__(*args, **kwargs)
self.tracker = Tracker(self)
self.no_track = []
self.commands = {
'ping': self.ping,
'stats': self.stats,
'track': self.track,
'help': self.help,
'chart': self.chart,
'age': self.age_users
}
try:
with open('DATA/USER_TRACKED.json', 'r') as f:
self.no_track = json.load(f)
except FileNotFoundError:
with open('DATA/USER_TRACKED.json', 'w') as f:
json.dump([], f)
async def on_ready(self):
print('Online now!')
await client.change_presence(activity=discord.Game(
name='@TrackerBot help'))
async def send(self):
async with aiohttp.ClientSession() as session:
async with session.post(
'https://api.fusiondiscordbots.com/{}/'.format(
self.user.id),
data={
'token': 'rv:J-HuSELAfy0pb',
'guilds': len(self.guilds),
'members': len([x for x in self.get_all_members()])
}) as resp:
print('returned {0.status} from api.fusiondiscordbots.com'.
format(resp))
async def on_guild_join(self, *args):
await self.send()
async def on_guild_remove(self, *args):
await self.send()
async def on_message(self, message):
if not await self.get_cmd(message):
pass
async def get_cmd(self, message):
if self.user.id in map(
lambda x: x.id,
message.mentions) and len(message.content.split(' ')) > 1:
if message.content.split(' ')[1] in self.commands.keys():
await self.commands[message.content.split(' ')[1]](message)
return True
return False
elif isinstance(message.channel, discord.DMChannel):
if message.content.split(' ')[0] in self.commands.keys():
await self.commands[message.content.split(' ')[0]](message)
return True
return False
async def ping(self, message):
t = message.created_at.timestamp()
e = await message.channel.send('pong')
delta = e.created_at.timestamp() - t
await e.edit(content='Pong! {}ms round trip'.format(round(delta *
1000)))
async def stats(self, message):
target = [
x for x in message.content.split(' ')[1:] if x.startswith('<@')
]
if len(target) > 0:
target = target[0][1:-1]
while target[0] not in '0123456789':
try:
target = target[1:]
except IndexError:
await message.channel.send(
'Error deciphering user tagged. Make sure it\'s formatted correctly!'
)
return
target = self.get_user(int(target))
if target is None:
await message.channel.send(
'Couldn\'t find user tagged. Are you sure they\'re real and a patron?'
)
return
else:
target = message.author
if self.tracker.getUser(target.id) is None:
return
em = discord.Embed(title='{}\'s stats'.format(target.name))
for key, data in self.tracker.getUser(target.id).items():
em.add_field(name=key,
value='{} minutes'.format(
round((data * self.tracker.INTERVAL) / 60)))
await message.channel.send(embed=em)
async def track(self, message):
if 'disable' in message.content:
await message.channel.send('Disabled tracking :thumbsup:')
self.no_track.append(message.author.id)
elif message.author.id in self.no_track:
await message.channel.send('Enabled tracking :thumbsup:')
self.no_track.remove(message.author.id)
else:
await message.channel.send(
'Tracking is currently enabled for you. Use `track disable` to disable tracking'
)
with open('DATA/USER_TRACKED.json', 'w') as f:
json.dump(self.no_track, f)
async def help(self, message):
await message.channel.send(embed=discord.Embed(description='''
`help` : Show this page
`stats [mention]` : Get online stats for a user
`track [disable]` : Enable or disable tracking for yourself
`chart [ignore="Offline,Online,GameName..."]` : Generate a pie chart of your activities
'''))
async def chart(self, message):
user = self.tracker.getUser(message.author.id)
ignore_list = []
listing = False
for i in message.content.split(' '):
if i.startswith('ignore="'):
ignore_list.append(i)
listing = True
elif listing:
ignore_list.append(i)
elif i[-1] == '"':
ignore_list.append(i)
listing = False
break
if len(ignore_list) > 0:
ignore_list = ' '.join(ignore_list)
listed = ignore_list.split('"')[1].replace('"', '').split(',')
listed = [i.strip().lower() for i in listed]
print(listed)
user = {k: v for k, v in user.items() if k.lower() not in listed}
pyplot.clf()
pyplot.axis('equal')
t = sum(user.values())
pyplot.pie(user.values(),
labels=user.keys(),
autopct=lambda x: '{}mins'.format(
int((((x * t) / 100) * self.tracker.INTERVAL) / 60)))
pyplot.savefig('curr.png')
f = open('curr.png', 'rb')
await message.channel.send(file=discord.File(f, 'chart.png'))
f.close()
async def Update(self):
await client.wait_until_ready()
while not client.is_closed():
await self.tracker.Update()
await asyncio.sleep(self.tracker.INTERVAL)
async def age_users(self, message):
users = [x for x in message.guild.members]
users.sort(key=lambda x: x.created_at.timestamp())
string = '\n'.join(
map(
lambda x: x.name + ' ' + x.created_at.strftime(
'%Y-%m-%d %H:%M:%S'), users))
new_str = ''
for i in string:
new_str += i
if len(new_str) > 1900:
await message.channel.send(new_str)
new_str = ''
await message.channel.send(new_str)
def get_patrons(self):
try:
p_server = self.get_guild(350391364896161793)
p_server2 = self.get_guild(366542432671760396)
p_roles = [
discord.utils.get(p_server.roles, name='Donor'),
discord.utils.get(p_server2.roles, name='Premium!')
]
premiums = [
user for user in itertools.chain(p_server.members,
p_server2.members)
if any([p in user.roles for p in p_roles])
]
return premiums
except:
return self.get_all_members()
class TorrentClient:
"""
TorrentClient is a local peer which is responsible for managing connections
to other peers to download and upload the data for the torrent.
It makes periodic calls to the tracker registered to the torrent meta-info to get peer
information for the torrent.
Each received peer is stored in the queue from which the PeerConnection objects consume
(maximum peer connections is defined by global MAX_PEER_CONNECTIONS variable).
"""
def __init__(self, torrent):
# Tracker object that defines methods to connect to the tracker
self.tracker = Tracker(torrent)
self.piece_manager = PieceManager(torrent)
# Queue of potential peers which the PeerConnection objects will consume
self.available_peers = asyncio.Queue()
# List of PeerConnection objects which might be connected to the peer.
# Else it is waiting to consume a peer from the available_peers queue
self.peers = []
self.abort = False
async def start(self):
"""
Start downloading the torrent data by connecting to the tracker and starting
the workers
"""
self.peers = [
PeerConnection(
self.available_peers,
self.tracker.torrent.info_hash,
self.tracker.peer_id,
self.piece_manager,
self._on_block_retrieved
)
for _ in range(MAX_PEER_CONNECTIONS)
]
# Timestamp of the last announce call to the tracker
previous_request = None
# Default request interval (in sec) to tracker
request_interval = 300 # sec
while True:
if self.abort:
logging.warning("Aborting download")
break
if self.piece_manager.complete:
logging.info("Download completed")
break
current_time = time.time()
# Check if request_interval time has passed so make announce call to tracker
if (not previous_request) or (previous_request + request_interval) < current_time:
tracker_response = await self.tracker.connect(
first=previous_request if previous_request else False,
uploaded=self.piece_manager.bytes_uploaded,
downloaded=self.piece_manager.bytes_downloaded
)
if tracker_response:
logging.debug("Got tracker response")
previous_request = current_time
request_interval = tracker_response.interval
self._empty_queue()
for peer in tracker_response.peers:
if peer:
self.available_peers.put_nowait(peer)
else:
logging.warning("Waiting for next tracker announce call, interval is {request_interval}".format(
request_interval=request_interval))
for peer in self.peers:
logging.debug("State of peer {id} is {status}".format(
id=peer.remote_id,
status=peer.my_state))
await asyncio.sleep(5)
self.stop()
def stop(self):
"""
Stop download and send stop signal to all peer objects
"""
self.abort = True
for peer in self.peers:
peer.stop()
self.piece_manager.close()
self.tracker.close()
def _empty_queue(self):
"""
Remove all peers from the queue
"""
while not self.available_peers.empty():
self.available_peers.get_nowait()
def _on_block_retrieved(self, peer_id, piece_index, block_offset, data):
"""
Callback function passed to PeerConnection object called when block is
retrieved from peer
:param peer_id: The id of peer the block was retrieved from
:param piece_index: The piece index of the block
:param block_offset: Block offset inside the piece
:param data: Binary block data
:return:
"""
self.piece_manager.block_received(
peer_id=peer_id,
piece_index=piece_index,
block_offset=block_offset,
data=data
)
from tracker import Tracker, get_model_base_path
if args.benchmark > 0:
model_base_path = get_model_base_path(args.model_dir)
im = cv2.imread(os.path.join(model_base_path, "benchmark.bin"),
cv2.IMREAD_COLOR)
results = []
for model_type in [3, 2, 1, 0, -1]:
tracker = Tracker(
224,
224,
threshold=0.1,
max_threads=args.max_threads,
max_faces=1,
discard_after=0,
scan_every=0,
silent=True,
model_type=model_type,
model_dir=args.model_dir,
no_gaze=(model_type < 0),
detection_threshold=0.1,
use_retinaface=0,
max_feature_updates=900,
static_model=True if args.no_3d_adapt == 1 else False)
tracker.detected = 1
tracker.faces = [(0, 0, 224, 224)]
total = 0.0
for i in range(100):
start = time.perf_counter()
r = tracker.predict(im)
total += time.perf_counter() - start
print(1. / (total / 100.))
def main():
api_id, api_hash, phone = extract_telegram_cfg(get_cfg())
tr = Tracker(api_id=api_id, api_hash=api_hash, phone=phone)
tr.start()
def pcd_callback(self, msg):
rospy.logwarn("Getting pcd at: %d.%09ds, (%d,%d)",
msg.header.stamp.secs, msg.header.stamp.nsecs,
msg.height, msg.width)
pcd_original = pointcloud2_to_xyz_array(msg)
pcd = PointCloud(pcd_original.T)
self.plane, pcd_inlier, pcd_outlier = self.estimate_plane(pcd)
transform_matrix, trans, rot, euler = get_transformation(
frame_from='/velodyne',
frame_to='/world',
time_from=msg.header.stamp,
time_to=msg.header.stamp,
static_frame='/world',
tf_listener=self.tf_listener,
tf_ros=self.tf_ros)
if not transform_matrix is None:
plane_world_param = np.matmul(
np.linalg.inv(transform_matrix).T,
np.array(
[[self.plane.a, self.plane.b, self.plane.c,
self.plane.d]]).T)
plane_world_param = plane_world_param / np.linalg.norm(
plane_world_param[0:3])
if self.plane_tracker is None:
self.plane_tracker = Tracker(msg.header.stamp,
plane_world_param)
else:
self.plane_tracker.predict(msg.header.stamp)
self.plane_tracker.update(plane_world_param)
print("plane_world:", plane_world_param.T)
print("plane_traker:", self.plane_tracker.filter.x_post.T)
# self.plane_world = Plane3D(plane_world_param[0,0], plane_world_param[1,0], plane_world_param[2,0], plane_world_param[3,0])
self.plane_world = Plane3D(self.plane_tracker.filter.x_post[0, 0],
self.plane_tracker.filter.x_post[1, 0],
self.plane_tracker.filter.x_post[2, 0],
self.plane_tracker.filter.x_post[3, 0])
center_pos = np.matmul(
transform_matrix,
np.array([[
10, 0, (-self.plane.a * 10 - self.plane.d) / self.plane.c,
1
]]).T)
center_pos = center_pos[0:3].flatten()
# normal = np.matmul( transform_matrix, np.array([[0., 0., 1., 0.]]).T)
# normal = normal[0:3]
normal = None
marker_array = self.create_and_publish_plane_markers(
self.plane_world,
frame_id='world',
center=center_pos,
normal=normal)
self.pub_plane_markers.publish(marker_array)
plane_msg = Plane()
plane_msg.coef[0], plane_msg.coef[1], plane_msg.coef[
2], plane_msg.coef[
3] = self.plane.a, self.plane.b, self.plane.c, self.plane.d
self.pub_plane.publish(plane_msg)
# pcd_msg_inlier = create_point_cloud(pcd_inlier.T, frame_id='velodyne')
# pcd_msg_outlier = create_point_cloud(pcd_outlier.T, frame_id='velodyne')
pcd_msg_inlier = xyz_array_to_pointcloud2(pcd_inlier.T,
stamp=msg.header.stamp,
frame_id='velodyne')
pcd_msg_outlier = xyz_array_to_pointcloud2(pcd_outlier.T,
stamp=msg.header.stamp,
frame_id='velodyne')
self.pub_pcd_inlier.publish(pcd_msg_inlier)
self.pub_pcd_outlier.publish(pcd_msg_outlier)
rospy.logwarn("Finished plane estimation")
def test_tag_reset():
t = Tracker(Board(5, 500, 500, 500, othelloAI()))
t.bd.generate_board()
t.bd.disks[1][1].halo_tag = True
t.tag_reset()
assert t.bd.disks[1][1].halo_tag is False
# window_name = [window for window in windows if key_word in window][0]
try:
# initialize the WindowCapture class
# wincap = WindowCapture(window_name)
cap = cv.VideoCapture('mario.mp4')
# load the trained models
neutral_b = cv.CascadeClassifier('cascade_models/cascade_copy.xml')
jab = cv.CascadeClassifier('cascade_models/jab.xml')
sheild = cv.CascadeClassifier('cascade_models/sheild.xml')
count = MoveCounter()
t = TimeTracker()
labels = MoveLabels()
tracker = Tracker()
# This stores the locations at which images are recognized.
move_locs = {labels.neutral_b: [], labels.jab: [], labels.shield: []}
mario_training_filter = HsvFilter(0, 5, 0, 179, 255, 255, 0, 17, 0, 0)
# initialize vision class
vision = Vision('')
while (True):
# get an updated image of the game
# USE THIS IF YOU ARE USING SCREENCAPTURE
# smash_screenshot = wincap.get_screenshot()
else:
# Default
target_pos = [0.5, 0.5]
except ValueError:
# Default
target_pos = [0.5, 0.5]
else:
# Default
target_pos = [0.5, 0.5]
else:
# Default
target_pos = [0.5, 0.5]
time_step = 0.0001
tracker = Tracker()
#matlab_port = MatlabPort()
pivot_threshold = 30
forward_speed = 80
navigator = Navigator()
controller = Controller(time_step, forward_speed, pivot_threshold)
target_heading = 0
my_pos = [0, 0]
motor_input = 3
timer = time.time()
while(not navigator.has_arrived()):
#!flask/bin/python
from flask import Flask, render_template, redirect, url_for, request, jsonify
from tracker import Tracker
from stravalib import Client
app = Flask(__name__)
app.config.from_envvar('APP_SETTINGS')
tracker = Tracker()
@app.route("/")
def login():
c = Client()
url = c.authorization_url(client_id=app.config['STRAVA_CLIENT_ID'],
redirect_uri=url_for('.logged_in',
_external=True),
approval_prompt='auto')
return render_template('login.html', authorize_url=url)
@app.route("/strava-oauth")
def logged_in():
"""
Method called by Strava (redirect) that includes parameters.
- state
- code
- error
"""
error = request.args.get('error')
state = request.args.get('state')
def test_constructor():
t = Tracker('Hi')
assert t.bd == 'Hi'
assert t.computer_moves == []
assert t.tag_count == 0
def process_img(img):
img = cv2.undistort(img, mtx, dist, None, mtx)
preprocess_image = np.zeros_like(img[:,:,0])
gradx = abs_sobel_thresh(img, orient='x', thresh=(12, 255))
grady = abs_sobel_thresh(img, orient='y', thresh=(25, 255))
c_binary = color_thresh(img, sthresh=(100, 255), vthresh=(50, 255))
preprocess_image[((gradx == 1) & (grady == 1) | (c_binary == 1))] = 255
img_size = (img.shape[1], img.shape[0])
bot_width = .76
mid_width = .08
height_pct = .62
bottom_trim = .935
src = np.float32([
[img.shape[1]*(.5-mid_width/2), img.shape[0]*height_pct],
[img.shape[1]*(.5+mid_width/2), img.shape[0]*height_pct],
[img.shape[1]*(.5+bot_width/2), img.shape[0]*bottom_trim],
[img.shape[1]*(.5-bot_width/2), img.shape[0]*bottom_trim],
])
offset = img.shape[1]*.25
dst = np.float32([
[offset, 0],
[img.shape[1]-offset, 0],
[img.shape[1]-offset, img.shape[0]],
[offset, img.shape[0]]
])
M = cv2.getPerspectiveTransform(src, dst)
Minv = cv2.getPerspectiveTransform(dst, src)
warped = cv2.warpPerspective(preprocess_image, M, img_size, flags=cv2.INTER_LINEAR)
window_width = 25
window_height = 80
curve_centers = Tracker(Mywindow_width=window_width, Mywindow_height=window_height, Mymargin=25, My_ym=10/720, My_xm=4/384, Mysmooth_factor=15)
window_centroids = curve_centers.find_window_centroids(warped)
l_points = np.zeros_like(warped)
r_points = np.zeros_like(warped)
leftx = []
rightx = []
for level in range(len(window_centroids)):
leftx.append(window_centroids[level][0])
rightx.append(window_centroids[level][1])
l_mask = window_mask(window_width, window_height, warped, window_centroids[level][0], level)
r_mask = window_mask(window_width, window_height, warped, window_centroids[level][1], level)
l_points[(l_points == 255) | (l_mask == 1)] = 255
r_points[(r_points == 255) | (r_mask == 1)] = 255
# Draw
# template = np.array(r_points+l_points, np.uint8)
# zero_channel=np.zeros_like(template)
# template = np.array(cv2.merge((zero_channel, template, zero_channel)), np.uint8)
# warpage = np.array(cv2.merge((warped, warped, warped)), np.uint8)
# result = cv2.addWeighted(warpage, 1, template, 0.5, 0.0)
# result = warped
yvals = range(0, warped.shape[0])
res_yvals = np.arange(warped.shape[0]-(window_height/2), 0, -window_height)
left_fit = np.polyfit(res_yvals, leftx, 2)
left_fitx = left_fit[0]*yvals*yvals + left_fit[1]*yvals + left_fit[2]
left_fitx = np.array(left_fitx, np.int32)
right_fit = np.polyfit(res_yvals, rightx, 2)
right_fitx = right_fit[0]*yvals*yvals + right_fit[1]*yvals + right_fit[2]
right_fitx = np.array(right_fitx, np.int32)
left_lane = np.array(list(zip(np.concatenate((left_fitx-window_width/2,left_fitx[::-1]+window_width/2), axis=0),
np.concatenate((yvals, yvals[::-1]), axis=0))), np.int32)
right_lane = np.array(list(zip(np.concatenate((right_fitx-window_width/2,right_fitx[::-1]+window_width/2), axis=0),
np.concatenate((yvals, yvals[::-1]), axis=0))), np.int32)
inner_lane = np.array(list(zip(np.concatenate((left_fitx+window_width/2,right_fitx[::-1]-window_width/2), axis=0),
np.concatenate((yvals, yvals[::-1]), axis=0))), np.int32)
road = np.zeros_like(img)
road_bkg = np.zeros_like(img)
cv2.fillPoly(road, [left_lane], color=[255, 0,0])
cv2.fillPoly(road, [right_lane], color=[0, 0, 255])
cv2.fillPoly(road, [inner_lane], color=[0, 255, 0])
cv2.fillPoly(road_bkg, [left_lane], color=[255, 255, 255])
cv2.fillPoly(road_bkg, [right_lane], color=[255, 255, 255])
road_warped = cv2.warpPerspective(road, Minv, img_size, flags=cv2.INTER_LINEAR)
road_warped_bkg = cv2.warpPerspective(road_bkg, Minv, img_size, flags=cv2.INTER_LINEAR)
base = cv2.addWeighted(img, 1.0, road_warped_bkg, -1.0, 0.0)
result = cv2.addWeighted(base, 1.0, road_warped, .7, 0.0)
ym_per_pix = curve_centers.ym_per_pix
xm_per_pix= curve_centers.xm_per_pix
curve_fit_cr = np.polyfit(np.array(res_yvals,np.float32)*ym_per_pix, np.array(leftx, np.float32)*xm_per_pix, 2)
curverad = ((1 + (2*curve_fit_cr[0]*yvals[-1]*ym_per_pix + curve_fit_cr[1])**2)**1.5) / np.absolute(2*curve_fit_cr[0])
camera_center = (left_fitx[-1] + right_fitx[-1])/2
center_diff = (camera_center-warped.shape[1]/2)*xm_per_pix
side_pos = 'left'
if center_diff <=0:
side_pos = 'right'
cv2.putText(result, 'Radius of Curvature = '+str(round(curverad, 3))+'(m)',(50,50),cv2.FONT_HERSHEY_SIMPLEX,1,(255,255,255),2)
cv2.putText(result, 'Vehicle is '+str(abs(round(center_diff, 3)))+'m '+side_pos+' of center',(50,100),cv2.FONT_HERSHEY_SIMPLEX,1,(255,255,255),2)
return result
def run_camera(input_shape, model):
num_classes = 21
conf_thresh = 0.5
bbox_util = BBoxUtility(num_classes)
vid = cv2.VideoCapture(0)
sleep(1.0)
# Compute aspect ratio of video
vidw = vid.get(cv2.CAP_PROP_FRAME_WIDTH)
vidh = vid.get(cv2.CAP_PROP_FRAME_HEIGHT)
trackers = Tracker()
while True:
ret, origin_image = vid.read()
frame = origin_image
if not ret:
print("Done!")
return None
im_size = (input_shape[0], input_shape[1])
resized = cv2.resize(frame, im_size)
rgb = cv2.cvtColor(resized, cv2.COLOR_BGR2RGB)
inputs = [image.img_to_array(rgb)]
tmp_inp = np.array(inputs)
x = preprocess_input(tmp_inp)
y = model.predict(x)
results = bbox_util.detection_out(y)
if len(results) > 0 and len(results[0]) > 0:
det_label = results[0][:, 0]
det_conf = results[0][:, 1]
det_xmin = results[0][:, 2]
det_ymin = results[0][:, 3]
det_xmax = results[0][:, 4]
det_ymax = results[0][:, 5]
top_indices = [i for i, conf in enumerate(det_conf) if conf >= conf_thresh]
top_conf = det_conf[top_indices]
top_label_indices = det_label[top_indices].tolist()
top_xmin = det_xmin[top_indices]
top_ymin = det_ymin[top_indices]
top_xmax = det_xmax[top_indices]
top_ymax = det_ymax[top_indices]
if 15 not in top_label_indices:
pass
else:
trackers.bbox = []
trackers.features_current = []
trackers.index = []
for i in range(top_conf.shape[0]):
class_num = int(top_label_indices[i])
if class_num == 15:
xmin = int(round((top_xmin[i] * vidw) * 0.9))
ymin = int(round((top_ymin[i] * vidh) * 0.9))
xmax = int(round((top_xmax[i] * vidw) * 1.1)) if int(round(
(top_xmax[i] * vidw)) * 1.1) <= vidw else int(round(
top_xmax[i] * vidw))
ymax = int(round((top_ymax[i] * vidh) * 1.1)) if int(round(
(top_ymax[i] * vidh) * 1.1)) <= vidh else int(round(top_ymax[i] * vidh))
trackers.bbox.append([xmin, ymin, xmax, ymax])
trackers.features_current.append(
Extract_feature(cv2.resize(frame[ymin:ymax, xmin:xmax, :], (32, 32))))
if trackers.features_previous is None:
trackers.index.append(i for i in range(len(trackers.bbox)))
for j in range(len(trackers.features_current)):
cv2.rectangle(frame, (int(trackers.bbox[j][0]), int(trackers.bbox[j][1])),
(int(trackers.bbox[j][2]), int(trackers.bbox[j][3])), (255, 0, 0), 2)
cv2.putText(frame, "person: {}".format(trackers.index[j] + 1),
(trackers.bbox[j][0] + 10, trackers.bbox[j][1] + 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 1)
else:
trackers.match()
trackers.update()
for j in range(len(trackers.features_current)):
cv2.rectangle(frame, (int(trackers.bbox[j][0]), int(trackers.bbox[j][1])),
(int(trackers.bbox[j][2]), int(trackers.bbox[j][3])), (255, 0, 0), 2)
cv2.putText(frame, "person: {}".format(trackers.index[j] + 1),
(trackers.bbox[j][0] + 10, trackers.bbox[j][1] + 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 1)
cv2.imshow('tracking', frame)
if cv2.waitKey(5) & 0xFF == ord('q'):
break
for w in self.arc_weights:
s+="w"
s+=str(w)
for b in self.biases:
s+="b"
s+=str(b)
for g in self.gains:
s+="g"
s+=str(g)
for t in self.time_constants:
s+="t"
s+=str(t)
s+=")"
return s
def __repr__(self):
return self.__str__()
#Develop a single parameter, from the binary list representing it to a float.
def dev_parameter(glist, start, stop, binlim, lim):
return ((int( "".join(glist[start:stop]), 2 )*1.0/binlim)*lim)
if __name__ == '__main__':
beer = Beer()
beer.development()
print beer
print beer.ann
print beer.ann.arcs
tracker = Tracker(beer.ann)
tracker.run()
def main():
### GET LIST OF INPUT FILES FROM STORAGE CONTAINER FOR SPARK TO READ
# Read from config files
storage_config = ConfigReader("config.cfg", "azure-storage").get_config()
db_config = ConfigReader("config.cfg", "postgres").get_config()
# Get Azure storage info from config
storage_acct_name = storage_config["account_name"]
storage_acct_access_key = storage_config["access_key"]
storage_container = storage_config["container_name"]
mount_root = storage_config["mount_root"]
# Set Spark Azure storage account and key
storage_acct_key_str = f"fs.azure.account.key.{storage_acct_name}.blob.core.windows.net"
# Strings for setting accessing filenames from storage
file_type = "txt"
input_dir = "data"
file_suffix = f".{file_type}"
suffix_len = len(file_suffix)
mount_base_path = f"{mount_root}/{storage_container}"
# Set up container client
blob_service_client = BlobServiceClient(account_url=f"https://{storage_acct_name}.blob.core.windows.net", \
credential=storage_acct_access_key)
container_client = blob_service_client.get_container_client(
storage_container)
# Get list of file names
blob_list = container_client.list_blobs(name_starts_with=input_dir)
txtfile_paths = [
blob.name for blob in blob_list
if blob.name[-suffix_len:] == file_suffix
]
txtfile_full_paths = [
f"{mount_base_path}/{file}" for file in txtfile_paths
]
### SET UP SPARK SESSION AND RUN THROUGH STEPS
# Start spark session and set up storage account key
spark = SparkSession.builder.getOrCreate()
spark.conf.set(storage_acct_key_str, storage_acct_access_key)
steps = []
step_1 = PipelineStep1("Step 1: Ingest",
spark=spark,
mount_base_path=mount_base_path,
input_path=",".join(txtfile_full_paths),
output_path=f"{mount_base_path}/ingested-data")
steps.append(step_1)
step_2 = PipelineStep2("Step 2: Preprocess",
spark=spark,
mount_base_path=mount_base_path,
input_path=step_1.output_path,
output_path=f"{mount_base_path}/preprocessed-data")
steps.append(step_2)
step_3 = PipelineStep3("Step 3: ETL",
spark=spark,
mount_base_path=mount_base_path,
input_path=step_2.output_path,
output_path=f"{mount_base_path}/ETL-output")
steps.append(step_3)
# Attempt each step and insert results into db table
job_ids = []
for step in steps:
tracker = Tracker(job_name=step.name, db_config=db_config)
job_ids.append(tracker.job_id)
try:
step.run()
tracker.update_job_status("Success")
except Exception as e:
print(e)
tracker.update_job_status("Failed")
import sys import os.path sys.path.append(os.path.join(os.path.dirname(__file__), '..')) import socket from tracker import Tracker import pysocket p = pysocket.Pysocket() t = Tracker((p.get_lan_ip(), 50000)) t.run()
def tracking(self,
trackers,
dboxes,
features,
appts,
fid,
mtracker_id,
dist_th=0.1,
iou_th=0.5,
dist_resume_th=0.1,
resume_times_th=10,
area_th=2,
iou_th_single=0.75,
no_detection_frames_th=50,
other_overlap_th=0.15,
is_oks=True):
ntrackers = trackers
nmtracker_id = mtracker_id
# Hungarian algorithm
# valid trackers number
vtn = 0
# key is valid trackers sequence id, value is all trackers sequence id
valid_id = {}
for i, tracker in enumerate(trackers):
if tracker.state == 0:
continue
valid_id[vtn] = i
vtn += 1
# valid tracker is existed
if vtn > 0:
dist_matrix = np.ones((len(dboxes), vtn))
iou_matrix = np.ones((len(dboxes), vtn))
oks_matrix = np.ones((len(dboxes), vtn))
for i, dbox in enumerate(dboxes):
feature = features[i]
dvtn = 0
for tracker in trackers:
if tracker.state == 0:
continue
iou = compute_iou(tracker.latest_box[:4], dbox[:4])
iou_matrix[i, dvtn] = iou
if is_oks:
# oks = oks_iou(np.array(tracker.pose_pts[:17]), np.array(appts[i][:17]),
# calc_area(tracker.latest_box), calc_area(dbox))
oks = oks_iou(np.array(tracker.pose_pts[:17]),
np.array(appts[i][:17]),
calc_area(tracker.latest_box),
calc_area(dbox),
in_vis_thre=0.05)
oks_matrix[i, dvtn] = oks
# calculate feature distance
distance = GetDis(tracker.mfeature, feature)
dist_matrix[i, dvtn] = distance
dvtn += 1
# Hungarian algorithm, dist matrix
row_ind, col_ind = linear_sum_assignment(dist_matrix)
rind = row_ind.tolist()
cind = col_ind.tolist()
# Hungarian algorithm, iou matrix
if is_oks:
iou_matrix = oks_matrix
row_ind, col_ind = linear_sum_assignment(iou_matrix, maximize=True)
# rind_iou = row_ind.tolist()
cind_iou = col_ind.tolist()
# # Hungarian algorithm, iou matrix
# row_ind, col_ind = linear_sum_assignment(oks_matrix, maximize=True)
# # rind_iou = row_ind.tolist()
# cind_oks = col_ind.tolist()
# determine which is dependent, which is not
# which box has been updated
dbox_markers = [0] * len(dboxes)
# which tracker has been updated
tracker_markers = [0] * len(trackers)
# which valid tracker has been updated
valid_tracker_markers = [0] * vtn
for i, ri in enumerate(rind):
ci = cind[i]
distance = dist_matrix[ri, ci]
# ri of rind and rind_iou is same
ci_iou = cind_iou[i]
iou = iou_matrix[ri, ci_iou]
# if distance < dist_th:
if ci == ci_iou and distance < dist_th * 1.5 and iou > iou_th / 2:
ti = valid_id[ci]
dbox = dboxes[ri]
feature = features[ri]
appt = appts[ri]
# dbox overlap with other tracker
is_other_overlap = False
ri_iou_list = iou_matrix[ri, :].tolist()
for idx, ri_iou in enumerate(ri_iou_list):
if idx == ci_iou:
continue
if ri_iou > other_overlap_th:
is_other_overlap = True
break
trackers[ti].update(dbox,
fid,
feature,
appt,
is_track=False,
is_other_overlap=is_other_overlap)
dbox_markers[ri] = 1
tracker_markers[ti] = 1
valid_tracker_markers[ci] = 1
continue
elif ci != ci_iou and iou > iou_th and dist_matrix[
ri, ci_iou] < dist_th:
ti = valid_id[ci_iou]
dbox = dboxes[ri]
feature = features[ri]
appt = appts[ri]
# dbox overlap with other tracker
is_other_overlap = False
ri_iou_list = iou_matrix[ri, :].tolist()
for idx, ri_iou in enumerate(ri_iou_list):
if idx == ci_iou:
continue
if ri_iou > other_overlap_th:
is_other_overlap = True
break
trackers[ti].update(dbox,
fid,
feature,
appt,
is_track=False,
is_other_overlap=is_other_overlap)
dbox_markers[ri] = 1
tracker_markers[ti] = 1
valid_tracker_markers[ci_iou] = 1
continue
elif ci != ci_iou and distance < dist_th and iou_matrix[
ri, ci] > iou_th:
ti = valid_id[ci]
dbox = dboxes[ri]
feature = features[ri]
appt = appts[ri]
# dbox overlap with other tracker
is_other_overlap = False
ri_iou_list = iou_matrix[ri, :].tolist()
for idx, ri_iou in enumerate(ri_iou_list):
if idx == ci:
continue
if ri_iou > other_overlap_th:
is_other_overlap = True
break
trackers[ti].update(dbox,
fid,
feature,
appt,
is_track=False,
is_other_overlap=is_other_overlap)
dbox_markers[ri] = 1
tracker_markers[ti] = 1
valid_tracker_markers[ci] = 1
continue
# add logic, try to match killed tracker
ktn = 0
# key is valid trackers sequence id, value is all trackers sequence id
killed_id = {}
for i, tracker in enumerate(trackers):
if tracker.state != 0 or (fid - tracker.latest_fid) > max(
50, self.drate * resume_times_th):
# if tracker.state != 0:
continue
killed_id[ktn] = i
ktn += 1
if dbox_markers.count(0) != 0 and ktn != 0:
dist_matrix_resume = np.ones((dbox_markers.count(0), ktn))
kdm = 0
# key is unmatched dbox idx, value is dboxes idx
unmatched_dbox_id = {}
for i, dm in enumerate(dbox_markers):
if dm != 0:
continue
unmatched_dbox_id[kdm] = i
dbox = dboxes[i]
# try to match
feature = features[i]
dktn = 0
for j, tracker in enumerate(trackers):
if tracker.state != 0 or (
fid - tracker.latest_fid) > max(
50, self.drate * resume_times_th):
# if tracker.state != 0:
continue
# calculate feature distance
distance = GetDis(tracker.mfeature, feature)
dist_matrix_resume[kdm, dktn] = distance
dktn += 1
kdm += 1
# Hungarian algorithm
row_ind, col_ind = linear_sum_assignment(dist_matrix_resume)
rind = row_ind.tolist()
cind = col_ind.tolist()
for i, ri in enumerate(rind):
ci = cind[i]
distance = dist_matrix_resume[ri, ci]
# similarity must be very high
if distance < dist_resume_th:
ti = killed_id[ci]
di = unmatched_dbox_id[ri]
dbox = dboxes[di]
feature = features[di]
appt = appts[di]
# area compare
darea = calc_area(dbox)
latest_box = trackers[ti].latest_box
larea = calc_area(latest_box)
# predict box compare
latest_fid = trackers[ti].latest_fid
delta_fid = fid - latest_fid
pred_velocity = trackers[ti].pred_velocity
vof = [
pred_velocity[0] * delta_fid,
pred_velocity[1] * delta_fid
]
pbox = [
int(latest_box[0] + vof[0]),
int(latest_box[1] + vof[1]),
int(latest_box[2] + vof[0]),
int(latest_box[3] + vof[1])
]
iou = compute_iou(dbox[:4], pbox)
# resume or not
if min(darea, larea) * area_th > max(
darea, larea) and iou > 0.001:
trackers[ti].resume()
trackers[ti].update(dbox,
fid,
feature,
appt,
is_track=False)
dbox_markers[di] = 1
tracker_markers[ti] = 1
# new tracker
for i, dm in enumerate(dbox_markers):
if dm != 0:
continue
# new tracker, using valid_tracker_markers info
iou_list = iou_matrix[i, :].tolist()
dist_list = dist_matrix[i, :].tolist()
mark = True
marked_vtm = {}
vcid = None
for cid, vtm in enumerate(valid_tracker_markers):
# tracker not matched before and iou is large
if vtm == 0 and (iou_list[cid] > iou_th or
(iou_list[cid] > 0.01
and dist_list[cid] < dist_th / 2)):
mark = False
vcid = cid
break
# iou area / dbox or iou area / tracker.latest_box is high
if vtm == 0:
ti = valid_id[cid]
dbox = dboxes[i]
latest_box = trackers[ti].latest_box
iou, area, darea, lbarea = compute_iou(
dbox[:4], latest_box[:4], True)
# if dist_list[cid] < dist_th and (area / darea > iou_th_single or area / lbarea > iou_th):
# seems overlap area in tracker cannot be too large, so using a smaller thresholld, as the detected box will be changed (part/full)
diou = area / darea
lbiou = area / lbarea
if diou > iou_th_single or lbiou > iou_th:
mark = False
# not creedy, otherwise mis-match
marked_vtm[cid] = diou + lbiou
if not mark:
if len(marked_vtm) != 0:
vcid = sorted(marked_vtm.items(),
key=lambda item: item[1],
reverse=True)[0][0]
ti = valid_id[vcid]
dbox = dboxes[i]
feature = features[i]
appt = appts[i]
# dbox overlap with other tracker
is_other_overlap = False
ri_iou_list = iou_list
for idx, ri_iou in enumerate(ri_iou_list):
if idx == ti:
continue
if ri_iou > other_overlap_th:
is_other_overlap = True
break
trackers[ti].update(dbox,
fid,
feature,
appt,
is_track=False,
is_other_overlap=is_other_overlap)
valid_tracker_markers[vcid] = 1
else:
# new tracker
nmtracker_id += 1
nt = Tracker(nmtracker_id)
dbox = dboxes[i]
feature = features[i]
appt = appts[i]
nt.update(dbox, fid, feature, appt, is_track=False)
ntrackers.append(nt)
# mark which tracker is not be detected
for i, tm in enumerate(tracker_markers):
if tm != 0 or ntrackers[i].state == 0:
continue
ntrackers[i].update_no_detection_times()
# ntrackers[i].update_no_detection_times(no_detection_frames_th // self.drate)
else:
# new tracker
for i, dbox in enumerate(dboxes):
nmtracker_id += 1
nt = Tracker(nmtracker_id)
feature = features[i]
appt = appts[i]
nt.update(dbox, fid, feature, appt, is_track=False)
ntrackers.append(nt)
return ntrackers, nmtracker_id