9397956920
update version number to 1.0.5
2019-10-22 23:06:25 +01:00
edf0b73e21
add OSNet-AIN code
2019-10-22 22:48:53 +01:00
fad38231c1
add radam optimizer
2019-10-22 22:47:08 +01:00
894de59bff
add new ref to osnet.py
2019-10-22 22:46:30 +01:00
1a5db72550
update annotation in RandomPatch
2019-10-22 22:45:48 +01:00
17887c05fd
version=1.0.3
2019-10-04 12:15:32 +01:00
cf553d367f
bug fix: pass video-specific args to dataset.__add__
2019-10-04 12:09:28 +01:00
7a73dca4ac
bug fix: img_path->path
2019-09-13 23:23:13 +01:00
0a60d2c6a7
version=1.0.2
2019-09-01 15:04:14 +01:00
0ec14d9afc
add random patch data augmentation
2019-09-01 15:03:42 +01:00
9131c537d5
[upgrade to 1.0.1] add cuhk02
2019-08-29 17:05:29 +01:00
5507b38c86
update v0.9.4 to v1.0.0
2019-08-26 13:26:10 +01:00
5365619606
add config
2019-08-26 10:34:31 +01:00
046ff13c07
minor change
2019-08-23 22:09:35 +01:00
d9cdb584e2
0.9.4
2019-08-23 22:05:09 +01:00
f4451e5ea9
fix visrank bug for video reid
2019-08-23 22:04:36 +01:00
2383698d52
Fix Visual Rank Video Reid Error
...
Fix video reid error while making sure it still works for image reid.
When trying to save visual ranks for video reid. It crashes on statement `qdir = osp.join(save_dir, osp.basename(osp.splitext(qimg_path)[0]))` that is because qimg_path can be a tuple or list for video reid. Hence, osp.splittext gives error when input is not a str. To fix this I make sure that osp.splitext always gets pass an str regardless if we doing video or image reid.
Let me know if you have any questions.
2019-08-23 12:12:32 -07:00
a23817173c
upgrade to 0.9.3
2019-08-22 23:32:23 +01:00
a40ba19207
add cosine lr
2019-08-22 23:17:47 +01:00
92f1ea215a
add tensorboard
2019-08-22 22:41:21 +01:00
1558389f37
osnet.py: rename residual->identity; engine: add num_batches=len(dloader)
2019-08-21 23:02:05 +01:00
a20f8a66bd
output_featuremaps_only -> return_featuremaps
2019-08-05 14:01:39 +01:00
d5863e0e82
0.9.1
2019-08-03 23:22:44 +01:00
1732d9e5da
0.9.0
2019-08-03 23:15:42 +01:00
e4fb577693
update docs for visrank and visactmap
2019-08-03 23:12:37 +01:00
84249ad412
[update visactmap] change channel order
2019-08-03 23:01:21 +01:00
d021fafff1
set visrank_topk=10 as default
2019-08-03 22:59:10 +01:00
28189cf1a6
[update visrank] use grid_img instead of plt.imshow
2019-08-03 22:30:22 +01:00
d72b85a1b8
set output_featuremaps_only=False in osnet
2019-08-03 21:52:52 +01:00
a08a85d8e4
[update visactmap] use cv2.imwrite instead of pyplot
2019-08-03 21:42:50 +01:00
7e0249a084
add output_featuremaps_only to foward() in osnet
2019-08-03 18:25:34 +01:00
3d64d9eaed
add default font size to visactmap
2019-08-03 18:07:51 +01:00
a6a5d39f68
print progress in visrank
2019-08-03 17:49:19 +01:00
27d4f2f093
add visactmap code; upgrade visrank to plot single figure
2019-08-03 17:35:06 +01:00
d398e66b37
allow empty transforms
2019-07-17 12:45:58 +01:00
1f891c1b36
[upgrade] 0.8.0=>0.8.1
2019-07-08 15:47:56 +01:00
9e418c0f5a
print resize(height, width)
2019-07-08 15:41:38 +01:00
d20b63c1a0
resize image to target size
2019-07-05 09:05:03 +01:00
87fd094a2c
Update datamanager.py
2019-07-04 16:39:11 +10:00
65a8f44d80
[datamanager] add comment
2019-07-03 15:54:20 +01:00
6829805723
[version] 0.7.8 => 0.8.0
2019-07-03 13:40:23 +01:00
60d27cf9e9
add osnet code
2019-07-03 13:37:25 +01:00
709018d948
add argument 'transforms' to datamanager
2019-07-03 12:05:55 +01:00
554807ad36
update mobilenetv2_1dot0 => mobilenetv2_x1_0 and the same for x1_4
2019-07-03 11:33:10 +01:00
fb4d585127
(epoch+1)>=start_eval
2019-06-13 23:11:07 +01:00
752cb82682
change attribute name: last_linear -> classifier
2019-06-03 09:26:04 +01:00
86ddbd97aa
update comment
2019-05-30 14:25:55 +01:00
250cd30ae2
0.7.8
2019-05-28 23:07:37 +01:00
9963cff629
update imagenet weights path
2019-05-28 15:33:04 +01:00
31068ca233
add global keyword
2019-05-24 17:13:30 +01:00
kaiyangzhou
KaiyangZhou
Abner Ayala-Acevedo
rick-hao