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)
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
seem ok	53fa176c31669a0e89b04adf290cb398f0316c45	Thai Thien	2019-08-24 18:26:31
flow ok	ad849681000818dfbcd0c1715c2858aed7236041	Thai Thien	2019-08-24 17:00:02
wip	23c3ec48497782bbc91d829e1c8a682502360ab9	Thai Thien	2019-08-24 14:19:22
work in progress, try to use https://pytorch.org/ignite/quickstart.html	39c824fe8fc2501628ee42c236a844df45521007	Thai Thien	2019-08-24 07:41:46

Commit a568fd7f294a8bd31b3db78437b4b6b51b5b41b9 - debug number not divisible by 8
Author: Thai Thien
Author date (UTC): 2019-09-22 04:36
Committer name: Thai Thien
Committer date (UTC): 2019-09-22 04:36
Parent(s): 967074890d14ab0eefc277801860270a468e8f9f
Signing key:
Tree: 5c73388ffc999be12841cbab2753755c7d4d7861

File	Lines added	Lines deleted
data_flow.py	2	2
main_pacnn.py	16	11
models/__init__.py	1	1
models/pacnn.py	8	1
models/test_PACNNWithPerspectiveMap.py	26	3

File data_flow.py changed (mode: 100644) (index 7e0162d..924b307)
...	...	def load_data_shanghaitech_pacnn(img_path, train=True):
98	98	target1 = cv2.resize(target, (int(target.shape[1] / 8), int(target.shape[0] / 8)),	target1 = cv2.resize(target, (int(target.shape[1] / 8), int(target.shape[0] / 8)),
99	99	interpolation=cv2.INTER_CUBIC) * 64	interpolation=cv2.INTER_CUBIC) * 64
100	100	target2 = cv2.resize(target, (int(target.shape[1] / 16), int(target.shape[0] / 16)),	target2 = cv2.resize(target, (int(target.shape[1] / 16), int(target.shape[0] / 16)),
101		interpolation=cv2.INTER_CUBIC) * 64 #*2
	101		interpolation=cv2.INTER_CUBIC) * 64 *2
102	102	target3 = cv2.resize(target, (int(target.shape[1] / 32), int(target.shape[0] / 32)),	target3 = cv2.resize(target, (int(target.shape[1] / 32), int(target.shape[0] / 32)),
103		interpolation=cv2.INTER_CUBIC) * 64 #*4
	103		interpolation=cv2.INTER_CUBIC) * 64 *4
104	104
105	105	return img, (target1, target2, target3)	return img, (target1, target2, target3)
106	106

File main_pacnn.py changed (mode: 100644) (index 9902b26..449320c)
...	...	from crowd_counting_error_metrics import CrowdCountingMeanAbsoluteError, CrowdCo
6	6	import torch	import torch
7	7	from torch import nn	from torch import nn
8	8	import torch.nn.functional as F	import torch.nn.functional as F
9		from models import CSRNet,PACNN
	9		from models import CSRNet, PACNN, PACNNWithPerspectiveMap
10	10	import os	import os
11	11	import cv2	import cv2
12	12	from torchvision import datasets, transforms	from torchvision import datasets, transforms

...	...	if __name__ == "__main__":
26	26	print(args)	print(args)
27	27	DATA_PATH = args.input	DATA_PATH = args.input
28	28	DATASET_NAME = "shanghaitech"	DATASET_NAME = "shanghaitech"
	29		PACNN_PERSPECTIVE_AWARE_MODEL = False
29	30
30	31	# create list	# create list
31	32	if DATASET_NAME is "shanghaitech":	if DATASET_NAME is "shanghaitech":

...	...	if __name__ == "__main__":
64	65	batch_size=1, num_workers=4)	batch_size=1, num_workers=4)
65	66
66	67	# create model	# create model
67		net = PACNN().to(device)
	68		net = PACNNWithPerspectiveMap(perspective_aware_mode=PACNN_PERSPECTIVE_AWARE_MODEL).to(device)
68	69	criterion_mse = nn.MSELoss(size_average=False).to(device)	criterion_mse = nn.MSELoss(size_average=False).to(device)
69		criterion_ssim = pytorch_ssim.SSIM(window_size=11).to(device)
	70		criterion_ssim = pytorch_ssim.SSIM(window_size=5).to(device)
70	71
71	72	optimizer = torch.optim.SGD(net.parameters(), args.lr,	optimizer = torch.optim.SGD(net.parameters(), args.lr,
72	73	momentum=args.momentum,	momentum=args.momentum,

...	...	if __name__ == "__main__":
84	85
85	86	# load data	# load data
86	87	d1_label, d2_label, d3_label = label	d1_label, d2_label, d3_label = label
87		d1_label = d1_label.to(device)
88		d2_label = d2_label.to(device)
89		d3_label = d3_label.to(device)
	88		d1_label = d1_label.to(device).unsqueeze(0)
	89		d2_label = d2_label.to(device).unsqueeze(0)
	90		d3_label = d3_label.to(device).unsqueeze(0)
90	91
91	92	# forward pass	# forward pass
92	93
93		d1, d2, d3 = net(train_img.to(device))
94		loss_1 = criterion_mse(d1, d1_label) + criterion_ssim(d1.unsqueeze(0), d1_label.unsqueeze(0))
95		loss_2 = criterion_mse(d2, d2_label) + criterion_ssim(d2.unsqueeze(0), d2_label.unsqueeze(0))
96		loss_3 = criterion_mse(d3, d3_label) + criterion_ssim(d3.unsqueeze(0), d3_label.unsqueeze(0))
97
	94		d1, d2, d3, p_s, p, d = net(train_img.to(device))
	95		loss_1 = criterion_mse(d1, d1_label) + criterion_ssim(d1, d1_label)
	96		loss_2 = criterion_mse(d2, d2_label) + criterion_ssim(d2, d2_label)
	97		loss_3 = criterion_mse(d3, d3_label) + criterion_ssim(d3, d3_label)
98	98	loss = loss_1 + loss_2 + loss_3	loss = loss_1 + loss_2 + loss_3
	99		if PACNN_PERSPECTIVE_AWARE_MODEL:
	100		# TODO: loss for perspective map here
	101		pass
	102		loss_d = criterion_mse(d, d1_label) + criterion_ssim(d, d1_label)
	103		loss += loss_d
99	104	loss.backward()	loss.backward()
100	105	optimizer.step()	optimizer.step()
101	106	loss_sum += loss.item()	loss_sum += loss.item()

File models/__init__.py changed (mode: 100644) (index 48fc576..7424e78)
1	1	from .csrnet import CSRNet	from .csrnet import CSRNet
2		from .pacnn import PACNN
	2		from .pacnn import PACNN, PACNNWithPerspectiveMap

File models/pacnn.py changed (mode: 100644) (index 82897d0..9e7203e)
...	...	class PACNNWithPerspectiveMap(nn.Module):
75	75	de23 = pespective_w_s * de2 + (1 - pespective_w_s)*(de2 + self.up23(de3))	de23 = pespective_w_s * de2 + (1 - pespective_w_s)*(de2 + self.up23(de3))
76	76	de = pespective_w * de1 + (1 - pespective_w)*(de1 + self.up12(de23))	de = pespective_w * de1 + (1 - pespective_w)*(de1 + self.up12(de23))
77	77	else:	else:
	78		#try:
	79		pespective_w_s = None
	80		pespective_w = None
78	81	de23 = (de2 + self.up23(de3))/2	de23 = (de2 + self.up23(de3))/2
79	82	de = (de1 + self.up12(de23))/2	de = (de1 + self.up12(de23))/2
80		return de
	83		# except Exception as e:
	84		# print("EXECEPTION ", e)
	85		# print(x.size())
	86		# print(de2.size(), de3.size())
	87		return de1, de2, de3, pespective_w_s, pespective_w, de
81	88
82	89	def count_param(net):	def count_param(net):
83	90	pytorch_total_params = sum(p.numel() for p in net.parameters())	pytorch_total_params = sum(p.numel() for p in net.parameters())

File models/test_PACNNWithPerspectiveMap.py changed (mode: 100644) (index cd3eab6..af8340f)
...	...	import torch
4	4
5	5	class TestPACNNWithPerspectiveMap(TestCase):	class TestPACNNWithPerspectiveMap(TestCase):
6	6
	7		def test_debug_avg_schema_pacnn(self):
	8		net = PACNNWithPerspectiveMap()
	9		image = torch.rand(1, 3, 330, 512)
	10		_, _, _, _, _, density_map = net(image)
	11		print(density_map.size())
	12
	13
7	14	def test_avg_schema_pacnn(self):	def test_avg_schema_pacnn(self):
8	15	net = PACNNWithPerspectiveMap()	net = PACNNWithPerspectiveMap()
9	16	# image	# image
10	17	# batch size, channel, h, w	# batch size, channel, h, w
11		image = torch.rand(1, 3, 224, 224)
12		density_map = net(image)
	18		image = torch.rand(1, 3, 330, 512)
	19		_, _, _, _, _, density_map = net(image)
	20		print(density_map.size())
	21		image2 = torch.rand(1, 3, 225, 225)
	22		_, _, _, _, _, density_map2 = net(image2)
	23		print(density_map2.size())
	24		image3 = torch.rand(1, 3, 226, 226)
	25		_, _, _, _, _, density_map3 = net(image3)
	26		print(density_map3.size())
	27
	28		image = torch.rand(1, 3, 227, 227)
	29		_, _, _, _, _, density_map = net(image)
13	30	print(density_map.size())	print(density_map.size())
	31		image2 = torch.rand(1, 3, 228, 228)
	32		_, _, _, _, _, density_map2 = net(image2)
	33		print(density_map2.size())
	34		image3 = torch.rand(1, 3, 229, 229)
	35		_, _, _, _, _, density_map3 = net(image3)
	36		print(density_map3.size())
14	37
15	38	def test_perspective_aware_schema_pacnn(self):	def test_perspective_aware_schema_pacnn(self):
16	39	net = PACNNWithPerspectiveMap(perspective_aware_mode=True)	net = PACNNWithPerspectiveMap(perspective_aware_mode=True)
17	40	# image	# image
18	41	# batch size, channel, h, w	# batch size, channel, h, w
19	42	image = torch.rand(1, 3, 224, 224)	image = torch.rand(1, 3, 224, 224)
20		density_map = net(image)
	43		_, _, _, _, _, density_map = net(image)
21	44	print(density_map.size())	print(density_map.size())

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