RocketGit

hahattpro / crowd_counting_framework (public) (License: Unspecified) (since 2020-12-23) (hash sha1)

backupfrom https://github.com/ttpro1995/crowd_counting_framework

Clone URLs: https://rocketgit.com/user/hahattpro/crowd_counting_framework ssh://rocketgit@ssh.rocketgit.com/user/hahattpro/crowd_counting_framework git://git.rocketgit.com/user/hahattpro/crowd_counting_framework

List of commits:

Subject	Hash	Author	Date (UTC)
apex not work so well	da8c0dd57297f972201f31d57e66897177922f48	Thai Thien	2019-09-24 17:25:59
fix data loader pacnn so it will scale up with correct number of people	11d55b50d764511f2491291f0208fee0905dec49	Thai Thien	2019-09-24 15:40:56
add comet ml	a9d4b89ce594f5e241168ccafdcdf0f150ea0ebb	Thai Thien	2019-09-23 17:07:58
fix pacnn avg schema	c2140a96886195782e5689c24aeeb4fe7a2db7ad	Thai Thien	2019-09-22 17:35:01
debug number not divisible by 8	a568fd7f294a8bd31b3db78437b4b6b51b5b41b9	Thai Thien	2019-09-22 04:36:06
pacnn	967074890d14ab0eefc277801860270a468e8f9f	Thai Thien	2019-09-22 03:54:48
wip: pacnn	2192d7c7b449fecf3868877d9cfbc09bb6f7ae98	Thai Thien	2019-09-22 03:44:56
wip: pacnn	37620e5a9bc0f9516ea964ec58d9bdaa1c40ff36	Thai Thien	2019-09-22 03:14:42
fix training flow	2b87b1b26c7296b64493fdc49fedb421b249dfa3	Thai Thien	2019-09-17 18:00:35
dataset script	bc5c052f5f956510ab95ef9a45434fd486c57fae	Thai Thien	2019-09-16 17:21:13
evaluator	ffc5bf8290ae0c469a9a18a2d061cfd1bfeee822	Thai Thien	2019-09-14 04:56:35
some more test for data loader	25173578cde7d4e9fe6c6140d1ee01caa4fcfc32	Thai Thien	2019-09-14 02:51:58
some visualize to debug data loader	e4f52007616acf307bddbde79c0fb4f8c649c785	Thai Thien	2019-09-13 17:35:45
wip	d7d44cad6774355bdfa45414258763f6c6a0c299	Thai Thien	2019-08-31 16:58:16
commit all	6dad7a58f7dbf9fc288ce9dd3e92be538851c2a7	Thai Thien	2019-08-29 19:10:44
input d1,d2,d3 match	fc2a809241f8b6356d964c63d40cbebd55ca5f6c	Thai Thien	2019-08-28 17:57:05
WIP	39eab26d061e61dfffbf164dbd5fd878299b7250	thient	2019-08-28 11:09:12
output of de is ok	dd770386674df3e0fbebafdfc48a9352bc28967d	thient	2019-08-28 10:54:09
code pacnn	c49537b5cc91e96e4e35c9338d2c95b9bb41c672	Thai Thien	2019-08-27 16:35:27
crowd counting stuff	da9f27a39cba9bdd021b6b5c562f5f7c2be50190	Thai Thien	2019-08-24 18:27:44

Commit da8c0dd57297f972201f31d57e66897177922f48 - apex not work so well
Author: Thai Thien
Author date (UTC): 2019-09-24 17:25
Committer name: Thai Thien
Committer date (UTC): 2019-09-24 17:25
Parent(s): 11d55b50d764511f2491291f0208fee0905dec49
Signing key:
Tree: 3bc1550981a3a195bfd04298691260aad4b656e4

File	Lines added	Lines deleted
main_pacnn.py	51	9

File main_pacnn.py changed (mode: 100644) (index a21c0f5..1078b73)
...	...	from evaluator import MAECalculator
18	18
19	19	from model_util import save_checkpoint	from model_util import save_checkpoint
20	20
	21		import apex
	22		from apex import amp
	23
21	24	if __name__ == "__main__":	if __name__ == "__main__":
22	25	# import comet_ml in the top of your file	# import comet_ml in the top of your file
23	26
24	27
25		MODEL_SAVE_NAME = "dev4"
	28		MODEL_SAVE_NAME = "dev5"
26	29	# Add the following code anywhere in your machine learning file	# Add the following code anywhere in your machine learning file
27	30	experiment = Experiment(api_key="S3mM1eMq6NumMxk2QJAXASkUM",	experiment = Experiment(api_key="S3mM1eMq6NumMxk2QJAXASkUM",
28	31	project_name="pacnn-dev", workspace="ttpro1995")	project_name="pacnn-dev", workspace="ttpro1995")

...	...	if __name__ == "__main__":
84	87	optimizer = torch.optim.SGD(net.parameters(), args.lr,	optimizer = torch.optim.SGD(net.parameters(), args.lr,
85	88	momentum=args.momentum,	momentum=args.momentum,
86	89	weight_decay=args.decay)	weight_decay=args.decay)
	90		# Allow Amp to perform casts as required by the opt_level
	91		net, optimizer = amp.initialize(net, optimizer, opt_level="O1", enabled=False)
	92
87	93	for e in range(10):	for e in range(10):
88	94	print("start epoch ", e)	print("start epoch ", e)
89	95	loss_sum = 0	loss_sum = 0

...	...	if __name__ == "__main__":
113	119	pass	pass
114	120	loss_d = criterion_mse(d, d1_label) + criterion_ssim(d, d1_label)	loss_d = criterion_mse(d, d1_label) + criterion_ssim(d, d1_label)
115	121	loss += loss_d	loss += loss_d
116		loss.backward()
	122		# loss.backward()
	123		with amp.scale_loss(loss, optimizer) as scaled_loss:
	124		scaled_loss.backward()
117	125	optimizer.step()	optimizer.step()
	126		optimizer.zero_grad()
118	127	loss_sum += loss.item()	loss_sum += loss.item()
119	128	sample += 1	sample += 1
120		optimizer.zero_grad()
121	129	counting += 1	counting += 1
122	130
123	131	if counting%10 ==0:	if counting%10 ==0:

...	...	if __name__ == "__main__":
135	143	print("=================================================================")	print("=================================================================")
136	144
137	145	save_checkpoint({	save_checkpoint({
138		'state_dict': net.state_dict(),
	146		'model': net.state_dict(),
	147		'optimizer': optimizer.state_dict(),
	148		# 'amp': amp.state_dict()
139	149	}, False, MODEL_SAVE_NAME)	}, False, MODEL_SAVE_NAME)
140	150
141	151
142
143		# evaluate
144
145
	152		# after epoch evaluate
	153		mae_calculator_d1 = MAECalculator()
	154		mae_calculator_d2 = MAECalculator()
	155		mae_calculator_d3 = MAECalculator()
	156		mae_calculator_final = MAECalculator()
	157		with torch.no_grad():
	158		for val_img, label in val_loader_pacnn:
	159		net.eval()
	160		# load data
	161		d1_label, d2_label, d3_label = label
	162
	163		# forward pass
	164		d1, d2, d3, p_s, p, d = net(val_img.to(device))
	165
	166		d1_label = d1_label.to(device)
	167		d2_label = d2_label.to(device)
	168		d3_label = d3_label.to(device)
	169
	170		# score
	171		mae_calculator_d1.eval(d1.cpu().detach().numpy(), d1_label.cpu().detach().numpy())
	172		mae_calculator_d2.eval(d2.cpu().detach().numpy(), d2_label.cpu().detach().numpy())
	173		mae_calculator_d3.eval(d3.cpu().detach().numpy(), d3_label.cpu().detach().numpy())
	174		mae_calculator_final.eval(d.cpu().detach().numpy(), d1_label.cpu().detach().numpy())
	175		print("count ", mae_calculator_d1.count)
	176		print("d1_val ", mae_calculator_d1.get_mae())
	177		print("d2_val ", mae_calculator_d2.get_mae())
	178		print("d3_val ", mae_calculator_d3.get_mae())
	179		print("dfinal_val ", mae_calculator_final.get_mae())
	180		experiment.log_metric("d1_val", mae_calculator_d1.get_mae())
	181		experiment.log_metric("d2_val", mae_calculator_d2.get_mae())
	182		experiment.log_metric("d3_val", mae_calculator_d3.get_mae())
	183		experiment.log_metric("dfinal_val", mae_calculator_final.get_mae())
	184
	185
	186		#############################################
	187		# done training evaluate
146	188	net = PACNNWithPerspectiveMap(PACNN_PERSPECTIVE_AWARE_MODEL).to(device)	net = PACNNWithPerspectiveMap(PACNN_PERSPECTIVE_AWARE_MODEL).to(device)
147	189	print(net)	print(net)
148	190
149	191	best_checkpoint = torch.load(MODEL_SAVE_NAME + "checkpoint.pth.tar")	best_checkpoint = torch.load(MODEL_SAVE_NAME + "checkpoint.pth.tar")
150		net.load_state_dict(best_checkpoint['state_dict'])
	192		net.load_state_dict(best_checkpoint['model'])
151	193
152	194	# device = "cpu"	# device = "cpu"
153	195	# TODO d1_val 155.97279205322266	# TODO d1_val 155.97279205322266

Hints:
Before first commit, do not forget to setup your git environment:

git config --global user.name "your_name_here"
git config --global user.email "your@email_here"

Clone this repository using HTTP(S):

git clone https://rocketgit.com/user/hahattpro/crowd_counting_framework

Clone this repository using ssh (do not forget to upload a key first):

git clone ssh://rocketgit@ssh.rocketgit.com/user/hahattpro/crowd_counting_framework

Clone this repository using git:

git clone git://git.rocketgit.com/user/hahattpro/crowd_counting_framework

You are allowed to anonymously push to this repository.
This means that your pushed commits will automatically be transformed into a merge request:

... clone the repository ...
... make some changes and some commits ...
git push origin main