def net_topology(self, topology):
self.mask = ma.Mask(self, topology).weights;
# this boolean specifies if the net has a fully connected topology
# (in this sense if it's not so the backprop and forward change a lot)
self.fully_connected = False;
class Pyro(Character):
#FIXME: width, height of pyro - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 198
maxhp = 120
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Flamethrower(game, state, self.id).id
class Engineer(Character):
# FIXME: width, height of engineer - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 180
maxhp = 120
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Shotgun(game, state, self.id).id
class Soldier(Character):
# FIXME: width, height of soldier - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 162
maxhp = 150
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Rocketlauncher(game, state, self.id).id
class Sniper(Character):
# FIXME: width, height of heavy - rectangle collision
# TODO: this class
collision_mask = mask.Mask(12, 33, True)
max_speed = 144
maxhp = 200
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Minigun(game, state, self.id).id
class Spy(Character):
# FIXME: width, height of spy - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 194.4
maxhp = 100
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Revolver(game, state, self.id).id
self.cloaking = False
class Medic(Character):
# FIXME: width, height of Medic - rectangle collision
# FIXME: offsets to be proper
collision_mask = mask.Mask(12, 33, True)
max_speed = 175
maxhp = 120
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Medigun(game, state, self.id).id
class Heavy(Character):
# FIXME: width, height of heavy - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 144
maxhp = 200
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Minigun(game, state, self.id).id
def step(self, game, state, frametime):
Character.step(self, game, state, frametime)
if self.get_player(state).leftmouse:
self.hspeed = min(54, max(-54, self.hspeed))
class Quote(Character):
# width, height of scout - rectangle collision
collision_mask = mask.Mask(12, 33, True)
max_speed = 252
maxhp = 100
run_power = 1.4;
def __init__(self, game, state, player_id):
Character.__init__(self, game, state, player_id)
self.hp = self.maxhp
self.weapon_id = weapon.Blade(game, state, self.id).id
self.can_doublejump = True
def jump(self, game, state):
if self.onground(game, state):
self.vspeed = -300
self.can_doublejump = True
elif self.can_doublejump:
self.vspeed = -300
self.can_doublejump = False
def find_n_dot_estimate(self, fix_mask):
'''
Estimate the number of dots and a mask assuming the chemical potential is mu_l everywhere
The mask is defined as a list of size len(x) where x is the x-grid. Each element is labelled as 'l0','l1' corresponding to the leads, 'di' corresponding
to the ith dot and 'bi' corresponding to the ith barrier.
Counting begins from 0.
'''
mu = self.mu_l[0]
mu_x = np.repeat(mu, len(self.V))
# Use mu - V ~ n to make a preliminary classfication
if (not fix_mask):
prelim_mask = mask.Mask(mu_x - self.V)
self.mask = prelim_mask
return prelim_mask.mask_info['num_dot']
else:
# return value from the old calculation
return self.mask.mask_info['num_dot']
import sys
import cv2
import utils
import detection
import mask
import sqlmanager
import schedule
import track
sys.path.insert(0, "analysis/")
import analyzer
#-----------------------------
#<------ Configuration ------>
#-----------------------------
mask = mask.Mask("config-files/maskConfig.json")
db = sqlmanager.SQLManager("config-files/MySQLConfig.json")
analyzerObjct = analyzer.loadAnalyzer("config-files/analyzerConfig.json")
#-----------------------------
#<---------- Main ----------->
#-----------------------------
if __name__ == "__main__":
# Load video here
cap = cv2.VideoCapture('videos/testRotonda.mp4')
# Should we resize the video frame?
# cap.set(3, 1280)
# cap.set(4, 720)
#%%
import datasets
import mask
import numpy as np
from scipy.stats import ttest_ind_from_stats
nii_prefix = 'mri/mwp1{}.nii'
centers_list = datasets.load_centers_mcad(use_nii=True, use_csv=False,
nii_prefix=nii_prefix)
_mask = mask.Mask('./data/mask', 'grey_matter_smoothed_005.nii')
d = _mask.get_mask_data().flatten()
#%%
def get_center_voxel_msn_by_label(center, label):
persons = center.get_by_label(label)
data = []
for person in persons:
data.append(np.asarray(person.nii.dataobj).flatten())
data = np.array(data)
mean = np.mean(data, axis=0)
std = np.std(data, axis=0)
count = data.shape[0]
return mean, std, count
def gen_voxel_msn(centers_list, label_eg, label_cg):
for center in centers_list:
mean_eg, std_eg, count_eg = get_center_voxel_msn_by_label(center, label_eg)
mean_cg, std_cg, count_cg = get_center_voxel_msn_by_label(center, label_cg)
if count_eg and count_cg:
t, p = ttest_ind_from_stats(mean_eg, std_eg, count_eg,
class Building_Sentry(entity.MovingObject):
max_hp = 100 # Maximum hitpoints the sentry can ever have
starting_hp = 25 # At what hitpoints the sentry will start building
collision_mask = mask.Mask(26, 19, True) # TODO: Implement changing masks
build_time = 2 # Number of secs it takes to build
hp_increment = (max_hp-starting_hp)/build_time
animation_increment = 10/build_time # 10 == number of frames in sentry build animation
def __init__(self, game, state, owner):
super(Building_Sentry, self).__init__(game, state)
self.hp = self.starting_hp
self.isfalling = True
self.animation_frame = 0
self.building_time = 0
self.owner_id = owner.id
self.x = owner.x
self.y = owner.y
self.team = owner.team
if owner.flip == True:
self.flip = True
else:
self.flip = False
def step(self, game, state, frametime):
if self.isfalling:
# If we've hit the floor, get us back out and build
while game.map.collision_mask.overlap(self.collision_mask, (int(round(self.x)), int(round(self.y)))):
self.y -= 1
self.isfalling = False
# Gravity
self.vspeed += 300 * frametime
# TODO: air resistance, not hard limit
self.vspeed = min(800, self.vspeed)
if not self.isfalling:
self.hspeed = 0
self.vspeed = 0
if self.hp <= 0:
self.destroy(state)
return
if self.building_time >= self.build_time:
# Cap hp at max hp
if self.hp >= self.max_hp:
self.hp = self.max_hp
# Create a finished sentry, and destroy the building sentry object
owner = state.entities[self.owner_id]
owner.sentry = Sentry(game, state, self.owner_id, self.x, self.y, self.hp, self.flip, self.team)
self.destroy(state)
else:
# Continue building
self.hp += self.hp_increment * frametime
self.building_time += frametime
self.animation_frame += self.animation_increment * frametime
def interpolate(self, prev_obj, next_obj, alpha):
super(Building_Sentry, self).interpolate(prev_obj, next_obj, alpha)
self.animation_frame = prev_obj.animation_frame + (next_obj.animation_frame - prev_obj.animation_frame) * alpha
self.hp = prev_obj.hp + (next_obj.hp - prev_obj.hp) * alpha
self.build_time = prev_obj.build_time + (next_obj.build_time - prev_obj.build_time) * alpha
def destroy(self, state):
# TODO: Sentry destruction syncing, bubble
super(Building_Sentry, self).destroy(state)
def file_mask_buf(self, file_path):
mask_ = mask.Mask()
mask_.masked(file_path)
def fileMask(self):
mask_ = mask.Mask()
mask_.masked(self.path_entry.get())
def main():
# Init logger
if not os.path.isdir(args.save_path):
os.makedirs(args.save_path)
if args.resume:
if not os.path.isdir(args.resume):
os.makedirs(args.resume)
log = open(os.path.join(args.save_path, '{}.txt'.format(args.description)), 'w')
print_log('save path : {}'.format(args.save_path), log)
state = {k: v for k, v in args._get_kwargs()}
print_log(state, log)
print_log("Random Seed: {}".format(args.manualSeed), log)
print_log("use cuda: {}".format(args.use_cuda), log)
print_log("python version : {}".format(sys.version.replace('\n', ' ')), log)
print_log("torch version : {}".format(torch.__version__), log)
print_log("cudnn version : {}".format(torch.backends.cudnn.version()), log)
print_log("Compress Rate: {}".format(args.rate), log)
print_log("Epoch prune: {}".format(args.epoch_prune), log)
print_log("description: {}".format(args.description), log)
# Init data loader
if args.dataset=='cifar10':
train_loader=dataset.cifar10DataLoader(True,args.batch_size,True,args.workers)
test_loader=dataset.cifar10DataLoader(False,args.batch_size,False,args.workers)
num_classes=10
elif args.dataset=='cifar100':
train_loader=dataset.cifar100DataLoader(True,args.batch_size,True,args.workers)
test_loader=dataset.cifar100DataLoader(False,args.batch_size,False,args.workers)
num_classes=100
elif args.dataset=='imagenet':
assert False,'Do not finish imagenet code'
else:
assert False,'Do not support dataset : {}'.format(args.dataset)
# Init model
if args.arch=='cifarvgg16':
net=models.vgg16_cifar(True,num_classes)
elif args.arch=='resnet32':
net=models.resnet32(num_classes)
elif args.arch=='resnet56':
net=models.resnet56(num_classes)
elif args.arch=='resnet110':
net=models.resnet110(num_classes)
else:
assert False,'Not finished'
print_log("=> network:\n {}".format(net),log)
net = torch.nn.DataParallel(net, device_ids=list(range(args.ngpu)))
# define loss function (criterion) and optimizer
criterion = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(net.parameters(), state['learning_rate'], momentum=state['momentum'],
weight_decay=state['decay'], nesterov=True)
if args.use_cuda:
net.cuda()
criterion.cuda()
recorder = RecorderMeter(args.epochs)
# optionally resume from a checkpoint
if args.resume:
if os.path.isfile(args.resume+'checkpoint.pth.tar'):
print_log("=> loading checkpoint '{}'".format(args.resume+'checkpoint.pth.tar'), log)
checkpoint = torch.load(args.resume+'checkpoint.pth.tar')
recorder = checkpoint['recorder']
args.start_epoch = checkpoint['epoch']
if args.use_state_dict:
net.load_state_dict(checkpoint['state_dict'])
else:
net = checkpoint['state_dict']
optimizer.load_state_dict(checkpoint['optimizer'])
print_log("=> loaded checkpoint '{}' (epoch {})".format(args.resume, checkpoint['epoch']), log)
if args.evaluate:
time1=time.time()
validate(test_loader,net,criterion,args.use_cuda,log)
time2=time.time()
print('validate function took %0.3f ms' % ((time2 - time1) * 1000.0))
return
else:
print_log("=> no checkpoint found at '{}'".format(args.resume), log)
else:
print_log("=> not use any checkpoint for {} model".format(args.description), log)
if args.original_train:
original_train.args.arch=args.arch
original_train.args.dataset=args.dataset
original_train.main()
return
comp_rate=args.rate
m=mask.Mask(net,args.use_cuda)
print("-" * 10 + "one epoch begin" + "-" * 10)
print("the compression rate now is %f" % comp_rate)
val_acc_1, val_los_1 = validate(test_loader, net, criterion, args.use_cuda,log)
print(" accu before is: %.3f %%" % val_acc_1)
m.model=net
print('before pruning')
m.init_mask(comp_rate,args.last_index)
m.do_mask()
print('after pruning')
m.print_weights_zero()
net=m.model#update net
if args.use_cuda:
net=net.cuda()
val_acc_2, val_los_2 = validate(test_loader, net, criterion, args.use_cuda,log)
print(" accu after is: %.3f %%" % val_acc_2)
#
start_time=time.time()
epoch_time=AverageMeter()
for epoch in range(args.start_epoch,args.epochs):
current_learning_rate=adjust_learning_rate(args.learning_rate,optimizer,epoch,args.gammas,args.schedule)
need_hour, need_mins, need_secs = convert_secs2time(epoch_time.avg * (args.epochs - epoch))
need_time = '[Need: {:02d}:{:02d}:{:02d}]'.format(need_hour, need_mins, need_secs)
print_log(
'\n==>>{:s} [Epoch={:03d}/{:03d}] {:s} [learning_rate={:6.4f}]'.format(time_string(), epoch, args.epochs,
need_time, current_learning_rate) \
+ ' [Best : Accuracy={:.2f}]'.format(recorder.max_accuracy(False)), log)
train_acc,train_los=train(train_loader,net,criterion,optimizer,epoch,args.use_cuda,log)
validate(test_loader, net, criterion,args.use_cuda, log)
if (epoch % args.epoch_prune == 0 or epoch == args.epochs - 1):
m.model=net
print('before pruning')
m.print_weights_zero()
m.init_mask(comp_rate,args.last_index)
m.do_mask()
print('after pruning')
m.print_weights_zero()
net=m.model
if args.use_cuda:
net=net.cuda()
val_acc_2, val_los_2 = validate(test_loader, net, criterion,args.use_cuda,log)
is_best = recorder.update(epoch, train_los, train_acc, val_los_2, val_acc_2)
if args.resume:
save_checkpoint({
'epoch': epoch + 1,
'state_dict': net,
'recorder': recorder,
'optimizer': optimizer.state_dict(),
}, is_best, args.resume, 'checkpoint.pth.tar')
print('save ckpt done')
epoch_time.update(time.time()-start_time)
start_time=time.time()
torch.save(net,args.model_save)
# torch.save(net,args.save_path)
flops.print_model_param_nums(net)
flops.count_model_param_flops(net,32,False)
log.close()
class Sentry(entity.MovingObject):
collision_mask = mask.Mask(26, 19, True)
def __init__(self, game, state, owner_id, x, y, hp, flip, team):
super(Sentry, self).__init__(game, state)
self.owner_id = owner_id
self.aiming_direction = 0
self.x = x
self.y = y
self.hp = hp
self.flip = flip
self.detection_radius = 375
self.team = team
self.rotating = False
self.turret_flip = flip
self.rotatestart = 0
self.rotateend = 4
self.rotateindex = self.rotatestart;
self.default_direction = 180 * self.flip
self.direction = self.default_direction
#targetting Queue
self.nearest_target = -1
self.target_queue = []
def step(self, game, state, frametime):
# TODO: Aim at nearest enemy
# TODO: Clean up
if self.hp <= 0:
self.destroy(state)
self.target_queue = [] #clear the list
for obj in state.entities.values():
if isinstance(obj, character.Character) and math.hypot(self.x-obj.x,self.y - obj.y) <= self.detection_radius:
target_tuple = (obj, math.hypot(self.x-obj.x,self.y - obj.y))
self.target_queue.append(target_tuple)
if len(self.target_queue) > 0: #TODO: implement point_direction and adjust priorities accordingly
self.target_queue.sort(key= lambda distance: distance[1]) #sort by the second item in the tuples; distance
self.nearest_target = self.target_queue[0][0] #get the first part of tuple
target_character = state.entities[self.nearest_target.id]
target_angle = function.point_direction(self.x,self.y,target_character.x,target_character.y)
self.direction = target_angle
if target_character.x > self.x and self.turret_flip == True:
self.rotating = True
elif target_character.x < self.x and self.turret_flip == False:
self.rotating = True
else:
self.nearest_target = -1
if self.nearest_target == -1 and self.flip != self.turret_flip: #reset to old position
self.rotating = True
self.direction = self.default_direction
if self.rotating == True:
self.rotateindex += 0.15
if (self.rotateindex >= self.rotateend):
self.rotating = False
self.turret_flip = not self.turret_flip
self.rotateindex = self.rotatestart
def interpolate(self, prev_obj, next_obj, alpha):
super(Sentry, self).interpolate(prev_obj, next_obj, alpha)
self.hp = prev_obj.hp + (next_obj.hp - prev_obj.hp) * alpha
self.direction = prev_obj.direction + (next_obj.direction - prev_obj.direction) * alpha
self.rotateindex = prev_obj.rotateindex + (next_obj.rotateindex - prev_obj.rotateindex) * alpha
if alpha < 0.5: self.rotating = prev_obj.rotating
else: self.rotating = next_obj.rotating
if alpha < 0.5: self.turret_flip = prev_obj.turret_flip
else: self.turret_flip = next_obj.turret_flip
def destroy(self, state):
# TODO: Sentry destruction syncing, bubble
super(Sentry, self).destroy(state)
owner = state.entities[self.owner_id]
owner.sentry = None