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)
print y and y_pred shape	90ab90465dabc4bd1171f4500eb01c45cca97420	Thai Thien	2020-12-06 15:23:09
ccnn baseline	f02c3084f28f811879e36e9a309993d468535dc3	Thai Thien	2020-12-06 15:08:11
ccnn shb and sha	d1d4152ada2c5dfc18c65299f32cb12935b90fe4	Thai Thien	2020-12-06 14:46:36
sha	3196ece275027e4d032a0bddb84ce310c15f2380	Thai Thien	2020-12-06 13:46:28
abs clamp	524c6f7196908527367ddcbfc340107e4092cc6e	Thai Thien	2020-12-06 11:16:12
use abs instead of ssim	92226d63c9732e66bf50f0b78e61f915ad2bce77	Thai Thien	2020-12-06 10:56:56
claim both y and y_pred	d1f6764a691fc2f2f1adc63ae65d63a8347d8a45	Thai Thien	2020-12-06 10:52:14
clamp 0.0	5569f7dad3c9e792e65636bb29b66a84ff250282	Thai Thien	2020-12-06 10:46:57
dataloader target crop	e41c17147de7c05f213329009c7662fbb2f1dc91	Thai Thien	2020-12-06 10:43:12
clamp min = 0	5245898e5282aa4cdc4b6539e68cb36a1c2d0c1f	Thai Thien	2020-12-06 10:40:47
fix metric	9739b99b836657377eb9dce05fafeef4a90546e9	Thai Thien	2020-12-06 10:38:37
eval density	d02232a419cd22bc79e0325c1c6791a7e5fc15b0	Thai Thien	2020-12-06 10:37:22
.	926498e13908770b072aeca0ccecbf5d3a64808e	Thai Thien	2020-12-06 10:27:13
if self.dataset_name == "shanghaitech_non_overlap_test_with_densitygt" and not self.train:	ca2bf6f9372ccdd5b880f2dd7be2023840afdd1d	Thai Thien	2020-12-06 10:24:21
yeah forgot to change ds name	bfdf3cfe27be5f51de300e231b6b79f903a71624	Thai Thien	2020-12-06 10:17:37
eval_only mode should not save	bf03ec80b43e070501c043f7428d55d8409b960c	Thai Thien	2020-12-06 10:14:22
now we output density map for testing	fb1f9f1c8abeb2311f0637db5f7b4c4ca4b43810	Thai Thien	2020-12-06 10:12:01
predict	8a312e6bcbe7a8b726cf1a01497db09bc21fea51	Thai Thien	2020-12-06 09:57:58
go ssim and psnr	d9b0d9c0417835c51ea59c66d85b8821480d0357	Thai Thien	2020-12-06 09:36:21
t4	624c9948c2099f16bd66f30a0c18093367b8a113	Thai Thien	2020-12-06 03:59:20

Commit 90ab90465dabc4bd1171f4500eb01c45cca97420 - print y and y_pred shape
Author: Thai Thien
Author date (UTC): 2020-12-06 15:23
Committer name: Thai Thien
Committer date (UTC): 2020-12-06 15:23
Parent(s): f02c3084f28f811879e36e9a309993d468535dc3
Signer:
Signing key:
Signing status: N
Tree: a0275017480c690418b991456bb720132963d5e7

File	Lines added	Lines deleted
crowd_counting_error_metrics.py	17	1
data_flow.py	5	5
playground/try_to_do_overlay.py	1	0

File crowd_counting_error_metrics.py changed (mode: 100644) (index f8d2d31..26be2d3)
1	1	from __future__ import division	from __future__ import division
2	2
3	3	import torch	import torch
	4		import torch.nn.functional as F
4	5	import math	import math
5	6	from ignite.exceptions import NotComputableError	from ignite.exceptions import NotComputableError
6	7	from ignite.metrics.metric import Metric	from ignite.metrics.metric import Metric

...	...	class CrowdCountingMeanSSIMabs(Metric):
129	130	# y = torch.clamp_min(y, min=0.0)	# y = torch.clamp_min(y, min=0.0)
130	131	y = torch.abs(y)	y = torch.abs(y)
131	132	y_pred = torch.abs(y_pred)	y_pred = torch.abs(y_pred)
	133		print("CrowdCountingMeanSSIMabs ")
	134		print("y_pred", y_pred.shape)
	135		print("y", y.shape)
	136
132	137
133	138	ssim_metric = piq.ssim(y, y_pred)	ssim_metric = piq.ssim(y, y_pred)
134	139

...	...	class CrowdCountingMeanPSNRabs(Metric):
161	166	# y = torch.clamp_min(y, min=0.0)	# y = torch.clamp_min(y, min=0.0)
162	167	y = torch.abs(y)	y = torch.abs(y)
163	168	y_pred = torch.abs(y_pred)	y_pred = torch.abs(y_pred)
	169		print("CrowdCountingMeanPSNRabs ")
	170		print("y_pred", y_pred.shape)
	171		print("y", y.shape)
	172
164	173	psnr_metric = piq.psnr(y, y_pred)	psnr_metric = piq.psnr(y, y_pred)
165	174
166	175
167	176
	177
168	178	self._sum += psnr_metric.item() * y.shape[0]	self._sum += psnr_metric.item() * y.shape[0]
169	179	# we multiply because ssim calculate mean of each image in batch	# we multiply because ssim calculate mean of each image in batch
170	180	# we multiply so we will divide correctly	# we multiply so we will divide correctly

...	...	class CrowdCountingMeanSSIMclamp(Metric):
194	204	y = output[1]	y = output[1]
195	205	y_pred = torch.clamp_min(y_pred, min=0.0)	y_pred = torch.clamp_min(y_pred, min=0.0)
196	206	y = torch.clamp_min(y, min=0.0)	y = torch.clamp_min(y, min=0.0)
197
	207		print("CrowdCountingMeanSSIMclamp ")
	208		print("y_pred", y_pred.shape)
	209		print("y", y.shape)
198	210
199	211	ssim_metric = piq.ssim(y, y_pred)	ssim_metric = piq.ssim(y, y_pred)
200	212

...	...	class CrowdCountingMeanPSNRclamp(Metric):
225	237	y_pred = torch.clamp_min(y_pred, min=0.0)	y_pred = torch.clamp_min(y_pred, min=0.0)
226	238	y = output[1]	y = output[1]
227	239	y = torch.clamp_min(y, min=0.0)	y = torch.clamp_min(y, min=0.0)
	240		print("CrowdCountingMeanPSNRclamp ")
	241		print("y_pred", y_pred.shape)
	242		print("y", y.shape)
	243
228	244
229	245	psnr_metric = piq.psnr(y, y_pred)	psnr_metric = piq.psnr(y, y_pred)
230	246

File data_flow.py changed (mode: 100644) (index 764d967..74ec54c)
...	...	def load_data_shanghaitech_non_overlap_test_with_densitygt(img_path, train=True,
657	657
658	658	gt_file = h5py.File(gt_path, 'r')	gt_file = h5py.File(gt_path, 'r')
659	659	target = np.asarray(gt_file['density'])	target = np.asarray(gt_file['density'])
660		target1 = cv2.resize(target,
661		(int(target.shape[1] / target_factor), int(target.shape[0] / target_factor)),
662		interpolation=cv2.INTER_CUBIC) * target_factor * target_factor
663		# target1 = target1.unsqueeze(0) # make dim (batch size, channel size, x, y) to make model output
664		target1 = np.expand_dims(target1,
	660		# target1 = cv2.resize(target,
	661		# (int(target.shape[1] / target_factor), int(target.shape[0] / target_factor)),
	662		# interpolation=cv2.INTER_CUBIC) * target_factor * target_factor
	663		# # target1 = target1.unsqueeze(0) # make dim (batch size, channel size, x, y) to make model output
	664		target1 = np.expand_dims(target,
665	665	axis=0) # make dim (batch size, channel size, x, y) to make model output	axis=0) # make dim (batch size, channel size, x, y) to make model output
666	666	return img_origin, target1	return img_origin, target1
667	667

File playground/try_to_do_overlay.py changed (mode: 100644) (index 187fcc6..61c2012)
...	...	print(img_tensor.shape)
13	13	print(density_map_tensor.shape)	print(density_map_tensor.shape)
14	14	print(density_map_tensor.sum())	print(density_map_tensor.sum())
15	15	density_map_tensor = torch.from_numpy(density_map_tensor).unsqueeze(dim=0).unsqueeze(dim=0)	density_map_tensor = torch.from_numpy(density_map_tensor).unsqueeze(dim=0).unsqueeze(dim=0)
	16		print("density_map_tensor.shape", density_map_tensor.shape) # torch.Size([1, 1, 46, 82])
16	17	# module = nn.UpsamplingBilinear2d(scale_factor=8)	# module = nn.UpsamplingBilinear2d(scale_factor=8)
17	18	# upsampling_density_map_tensor = module(density_map_tensor)	# upsampling_density_map_tensor = module(density_map_tensor)
18	19	upsampling_density_map_tensor = nn.functional.interpolate(density_map_tensor, scale_factor=8)/64	upsampling_density_map_tensor = nn.functional.interpolate(density_map_tensor, scale_factor=8)/64

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