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)
M4	c960a8e3ddbfb7fc57f3f843fa4184c063cf8cdb	Thai Thien	2020-04-16 14:22:37
typo again	3dbe3ce4634b8d4ca30b012851c5b9690b1d88d7	Thai Thien	2020-04-13 15:49:23
typo	9b9a84ed5bfffc6e8979fe8b9aa2d6411bfd70c2	Thai Thien	2020-04-13 15:47:51
typo	69cdd6f3037ef0357783ad4c3f8cdf8de2258c3b	Thai Thien	2020-04-13 15:38:14
small tall on split branch	36c80eee740df7449c112f4dd4925e0ffbc7ac5a	Thai Thien	2020-04-13 15:23:33
fix load model	68c03563a5aa3acda165eae104ff4d2df83201b2	Thai Thien	2020-04-09 17:30:14
get lr and weight_decay	b43666da2cb8bb5710f30d0f3bfd2c3b1e7b6473	Thai Thien	2020-04-09 17:26:13
1201 epoch (because cur epoch continue when load)	ee303fe3945fcc6cc3c26601039de77a58fa3d60	Thai Thien	2020-04-09 17:23:18
shb load model	062126f959c021577dbf08224aeb442ca308587c	Thai Thien	2020-04-09 17:20:27
4	2f49bfa380e997c177af1de34c8d6882ed7099e9	Thai Thien	2020-04-09 17:06:28
11	331644c623b1bf5a34c4db432e1526f6bc34a398	Thai Thien	2020-04-09 17:04:49
batchsize 24	f6aeba845ee1915cb0aeb8fce298e56a6ba40b3a	Thai Thien	2020-04-09 16:46:25
t10	91d9e83c80a6a533535fd91278b9175c839c9715	Thai Thien	2020-04-09 16:43:38
t9	927b97f2f285000ce7b407496c52da2c61539cb8	Thai Thien	2020-04-09 16:37:39
minor bugfix	cbab75b39a2d3495c1d07a6f5c127ccc6e7cfbf5	Thai Thien	2020-04-09 16:15:24
ccnn v2 t89 shb	15fca0fdfa56fb6abe4928642fbaf8b444fa5fbb	Thai Thien	2020-04-09 15:46:00
fixed: not discard under-size in 256	8a0ef0a9bfb0801933dd4403685fe65ebb5dac6a	Thai Thien	2020-04-08 16:10:46
data_flow fixed 256, we ignore too small image in Batch for now	34a56f59674dee5267dba356b3ebe469baa91baf	Thai Thien	2020-04-08 15:46:34
fix typo	e9e5fe46043233c703e0a872afd1510434d4b5b0	Thai Thien	2020-04-07 18:49:20
tryting to reproduce ccnn again	b3681bbf85bb50eb5430581d8779415520d78dc2	Thai Thien	2020-04-07 18:25:30

Commit c960a8e3ddbfb7fc57f3f843fa4184c063cf8cdb - M4
Author: Thai Thien
Author date (UTC): 2020-04-16 14:22
Committer name: Thai Thien
Committer date (UTC): 2020-04-16 14:22
Parent(s): 3dbe3ce4634b8d4ca30b012851c5b9690b1d88d7
Signing key:
Tree: 2747633767badc78982fc57c348ea760f769a9b4

File	Lines added	Lines deleted
experiment_meow_main.py	3	1
models/meow_experiment/kitten_meow_1.py	71	0
train_script/meow_one/M4_t1_sha.sh	4	4
train_script/meow_one/M4_t1_shb.sh	4	4
train_script/meow_one/big_tail/bigtail3_t1_sha_shb.sh	5	4
train_script/meow_one/big_tail/bigtail3_t1_shb_sha.sh	5	4

File experiment_meow_main.py changed (mode: 100644) (index 19ff638..84b1083)
...	...	from visualize_util import get_readable_time
10	10
11	11	import torch	import torch
12	12	from torch import nn	from torch import nn
13		from models.meow_experiment.kitten_meow_1 import M1, M2, M3
	13		from models.meow_experiment.kitten_meow_1 import M1, M2, M3, M4
14	14	from models.meow_experiment.ccnn_tail import BigTailM1, BigTailM2, BigTail3	from models.meow_experiment.ccnn_tail import BigTailM1, BigTailM2, BigTail3
15	15	from models import CustomCNNv2	from models import CustomCNNv2
16	16	import os	import os

...	...	if __name__ == "__main__":
66	66	model = M2()	model = M2()
67	67	elif model_name == "M3":	elif model_name == "M3":
68	68	model = M3()	model = M3()
	69		elif model_name == "M4":
	70		model = M4()
69	71	elif model_name == "CustomCNNv2":	elif model_name == "CustomCNNv2":
70	72	model = CustomCNNv2()	model = CustomCNNv2()
71	73	elif model_name == "BigTailM1":	elif model_name == "BigTailM1":

File models/meow_experiment/kitten_meow_1.py changed (mode: 100644) (index 43169dd..2daba3f)
...	...	class M2(nn.Module):
186	186
187	187	x = self.output(x)	x = self.output(x)
188	188	return x	return x
	189
	190
	191		class M4(nn.Module):
	192		"""
	193		A REAL-TIME DEEP NETWORK FOR CROWD COUNTING
	194		https://arxiv.org/pdf/2002.06515.pdf
	195		the improve version
	196
	197		we change 5x5 7x7 9x9 with 3x3
	198
	199		change tail to dilated max 60
	200		"""
	201		def __init__(self, load_weights=False):
	202		super(M4, self).__init__()
	203		self.model_note = "We replace 5x5 7x7 9x9 with 3x3, no batchnorm yet, change tail to dilated max 60 with dilated 2"
	204		# self.red_cnn = nn.Conv2d(3, 10, 9, padding=4)
	205		# self.green_cnn = nn.Conv2d(3, 14, 7, padding=3)
	206		# self.blue_cnn = nn.Conv2d(3, 16, 5, padding=2)
	207
	208		# ideal from crowd counting using DMCNN
	209		self.front_cnn_1 = nn.Conv2d(3, 20, 3, padding=1)
	210		self.front_cnn_2 = nn.Conv2d(20, 16, 3, padding=1)
	211		self.front_cnn_3 = nn.Conv2d(16, 14, 3, padding=1)
	212		self.front_cnn_4 = nn.Conv2d(14, 10, 3, padding=1)
	213		self.max_pooling = nn.MaxPool2d(kernel_size=2, stride=2)
	214
	215		self.c0 = nn.Conv2d(40, 60, 3, padding=2, dilation=2)
	216		self.c1 = nn.Conv2d(60, 60, 3, padding=2, dilation=2)
	217		self.c2 = nn.Conv2d(60, 60, 3, padding=2, dilation=2)
	218		self.c3 = nn.Conv2d(60, 30, 3, padding=2, dilation=2)
	219		self.c4 = nn.Conv2d(30, 15, 3, padding=2, dilation=2)
	220		self.c5 = nn.Conv2d(15, 10, 3, padding=2, dilation=2)
	221		self.output = nn.Conv2d(10, 1, 1)
	222
	223		def forward(self,x):
	224		#x_red = self.max_pooling(F.relu(self.red_cnn(x), inplace=True))
	225		#x_green = self.max_pooling(F.relu(self.green_cnn(x), inplace=True))
	226		#x_blue = self.max_pooling(F.relu(self.blue_cnn(x), inplace=True))
	227
	228		x_red = F.relu(self.front_cnn_1(x), inplace=True)
	229		x_red = F.relu(self.front_cnn_2(x_red), inplace=True)
	230		x_red = F.relu(self.front_cnn_3(x_red), inplace=True)
	231		x_red = F.relu(self.front_cnn_4(x_red), inplace=True)
	232		x_red = self.max_pooling(x_red)
	233
	234		x_green = F.relu(self.front_cnn_1(x), inplace=True)
	235		x_green = F.relu(self.front_cnn_2(x_green), inplace=True)
	236		x_green = F.relu(self.front_cnn_3(x_green), inplace=True)
	237		x_green = self.max_pooling(x_green)
	238
	239		x_blue = F.relu(self.front_cnn_1(x), inplace=True)
	240		x_blue = F.relu(self.front_cnn_2(x_blue), inplace=True)
	241		x_blue = self.max_pooling(x_blue)
	242
	243		x = torch.cat((x_red, x_green, x_blue), 1)
	244		x = F.relu(self.c0(x), inplace=True)
	245
	246		x = F.relu(self.c1(x), inplace=True)
	247
	248		x = F.relu(self.c2(x), inplace=True)
	249		x = self.max_pooling(x)
	250
	251		x = F.relu(self.c3(x), inplace=True)
	252		x = self.max_pooling(x)
	253
	254		x = F.relu(self.c4(x), inplace=True)
	255
	256		x = F.relu(self.c5(x), inplace=True)
	257
	258		x = self.output(x)
	259		return x

File train_script/meow_one/M4_t1_sha.sh copied from file train_script/meow_one/big_tail/bigtail3_t1_sha.sh (similarity 59%) (mode: 100644) (index fb030f2..1f07f29)
1		task="bigtail3_t1_sha"
	1		task="M4_t1_sha"
2	2
3		CUDA_VISIBLE_DEVICES=2 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
	3		CUDA_VISIBLE_DEVICES=3 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
4	4	--task_id $task \	--task_id $task \
5		--note "bigtail3 shanghaitech_rnd" \
	5		--note "M4 shanghaitech_rnd" \
6	6	--model "BigTail3" \	--model "BigTail3" \
7	7	--input /data/rnd/thient/thient_data/ShanghaiTech/part_A \	--input /data/rnd/thient/thient_data/ShanghaiTech/part_A \
8	8	--lr 1e-4 \	--lr 1e-4 \
9	9	--decay 1e-4 \	--decay 1e-4 \
10	10	--datasetname shanghaitech_20p \	--datasetname shanghaitech_20p \
11		--epochs 602 > logs/$task.log &
	11		--epochs 301 > logs/$task.log &
12	12
13	13	echo logs/$task.log # for convenience	echo logs/$task.log # for convenience

File train_script/meow_one/M4_t1_shb.sh copied from file train_script/meow_one/big_tail/bigtail3_t2_shb.sh (similarity 77%) (mode: 100644) (index ae56376..ecde449)
1		task="bigtail3_t2_shb"
	1		task="M4_t1_shb"
2	2
3	3	CUDA_VISIBLE_DEVICES=3 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \	CUDA_VISIBLE_DEVICES=3 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
4	4	--task_id $task \	--task_id $task \
5		--note "bigtail3 shanghaitech_rnd e-5 lr" \
	5		--note "M4 shanghaitech_rnd" \
6	6	--model "BigTail3" \	--model "BigTail3" \
7	7	--input /data/rnd/thient/thient_data/ShanghaiTech/part_B \	--input /data/rnd/thient/thient_data/ShanghaiTech/part_B \
8		--lr 1e-5 \
9		--decay 1e-5 \
	8		--lr 1e-4 \
	9		--decay 1e-4 \
10	10	--batch_size 8 \	--batch_size 8 \
11	11	--datasetname shanghaitech_rnd \	--datasetname shanghaitech_rnd \
12	12	--epochs 301 > logs/$task.log &	--epochs 301 > logs/$task.log &

File train_script/meow_one/big_tail/bigtail3_t1_sha_shb.sh copied from file train_script/meow_one/big_tail/bigtail3_t1_shb.sh (similarity 57%) (mode: 100644) (index 30e7fac..f1013ef)
1		task="bigtail3_t1_shb"
	1		task="bigtail3_t1_sha_shb"
2	2
3		CUDA_VISIBLE_DEVICES=1 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
	3		CUDA_VISIBLE_DEVICES=2 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
4	4	--task_id $task \	--task_id $task \
5		--note "bigtail3 shanghaitech_rnd" \
	5		--note "bigtail3 sha 20p then shb rnd crop" \
6	6	--model "BigTail3" \	--model "BigTail3" \
7	7	--input /data/rnd/thient/thient_data/ShanghaiTech/part_B \	--input /data/rnd/thient/thient_data/ShanghaiTech/part_B \
8	8	--lr 1e-4 \	--lr 1e-4 \
9	9	--decay 1e-4 \	--decay 1e-4 \
10	10	--batch_size 8 \	--batch_size 8 \
	11		--load_model \
11	12	--datasetname shanghaitech_rnd \	--datasetname shanghaitech_rnd \
12		--epochs 301 > logs/$task.log &
	13		--epochs 302 > logs/$task.log &
13	14
14	15	echo logs/$task.log # for convenience	echo logs/$task.log # for convenience

File train_script/meow_one/big_tail/bigtail3_t1_shb_sha.sh copied from file train_script/meow_one/big_tail/bigtail3_t1_sha.sh (similarity 54%) (mode: 100644) (index fb030f2..95f7218)
1		task="bigtail3_t1_sha"
	1		task="bigtail3_t1_shb_sha"
2	2
3		CUDA_VISIBLE_DEVICES=2 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
	3		CUDA_VISIBLE_DEVICES=1 HTTPS_PROXY="http://10.60.28.99:86" nohup python experiment_meow_main.py \
4	4	--task_id $task \	--task_id $task \
5		--note "bigtail3 shanghaitech_rnd" \
	5		--note "bigtail3 shanghaitech_rnd then train sha 20p" \
6	6	--model "BigTail3" \	--model "BigTail3" \
7	7	--input /data/rnd/thient/thient_data/ShanghaiTech/part_A \	--input /data/rnd/thient/thient_data/ShanghaiTech/part_A \
8	8	--lr 1e-4 \	--lr 1e-4 \
9	9	--decay 1e-4 \	--decay 1e-4 \
	10		--load_model \
10	11	--datasetname shanghaitech_20p \	--datasetname shanghaitech_20p \
11		--epochs 602 > logs/$task.log &
	12		--epochs 601 > logs/$task.log &
12	13
13	14	echo logs/$task.log # for convenience	echo logs/$task.log # for convenience

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