OrgLion-ML
/
deep-text-recognition-benchmark
mirror of https://github.com/clovaai/deep-text-recognition-benchmark.git
1
0
Fork 0
Code Issues Releases Wiki Activity
99 Commits 1 Branch 0 Tags 29 MiB
Commit Graph

99 Commits

Author SHA1 Message Date
Baek JeongHun 1493d2e8cd
Update README.md 2019-10-22 21:01:25 +09:00
Baek JeongHun e8b465d477 add confidence score 2019-10-22 11:46:45 +00:00
Baek JeongHun 4381a5fb46
Merge pull request #91 from edwardpwtsoi/master 
updated grid_sample arg align_corners=True
 2019-10-22 14:31:26 +09:00
edwardpwtsoi f5cf8555ef
updated grid_sample arg align_corners=True 
UserWarning: Default grid_sample and affine_grid behavior will be changed to align_corners=False from 1.4.0. See the documentation of grid_sample for details.
 2019-10-13 20:57:50 +08:00
Baek JeongHun ff7f7dbd71
Update README.md 2019-10-10 17:16:32 +09:00
Baek JeongHun 6dc16df598 (ctc_a) (ctc_b) and fine tuning 2019-10-07 05:33:44 +00:00
Baek JeongHun e34c99c386
Update README.md 2019-10-03 16:59:42 +09:00
Baek JeongHun 7a8717eb50
Update README.md 2019-10-03 15:41:40 +09:00
Baek JeongHun f2c54ae2a4
Delete trade-off.jpg 2019-10-03 15:22:38 +09:00
Baek JeongHun 464b3395df
Update README.md 2019-10-03 15:21:42 +09:00
Baek JeongHun 6985b0fa0a
Add files via upload 2019-10-03 15:21:03 +09:00
Baek JeongHun 1d85fca2c3
Update README.md 2019-10-01 12:02:39 +09:00
Baek JeongHun 235095fc9a
Update transformation.py 2019-09-27 18:06:16 +09:00
Baek JeongHun fc62d2232d 'minor' 2019-08-10 18:29:40 +09:00
Baek JeongHun 1c6efa5218 'fix_ctc_loss_issue' 2019-08-07 16:42:42 +09:00
Baek JeongHun 8f7255fb4b .cuda() to .to(device) 2019-08-03 08:03:46 +00:00
Baek JeongHun 7498f48452 no_grad instead of requires_grad False 2019-08-03 06:55:32 +00:00
Baek JeongHun d62daf3baa
Merge pull request #48 from YacobBY/patch-2 
replaced deprecated fromstring() with frombuffer()
 2019-08-03 15:04:38 +09:00
Yacob cbc1d9df0f
replaced deprecated fromstring() with frombuffer() 
Kept seeing

`DeprecationWarning: The binary mode of fromstring is deprecated, as it behaves surprisingly on unicode inputs. Use frombuffer instead imageBuf = np.fromstring(imageBin, dtype=np.uint8)`, this fixes the issue.
 2019-08-01 16:28:41 +02:00
Baek JeongHun 9128831f7b
Update README.md 2019-08-01 12:03:40 +09:00
Baek JeongHun b3b3889295
Merge pull request #46 from coallaoh/patch-1 
Update README.md
 2019-08-01 12:03:07 +09:00
Seong Joon Oh 42d9506787
Update README.md 2019-08-01 08:26:31 +09:00
Baek JeongHun 533fdf01ba fix encode with batch_max_length 2019-07-31 08:04:56 +00:00
Baek JeongHun bda81065a5 fix encode with batch_max_length 2019-07-31 08:03:53 +00:00
Baek JeongHun a239f46df3 'fix' 2019-07-26 15:02:05 +09:00
Baek JeongHun c189d3d8dc
Update README.md 2019-07-25 01:27:24 +09:00
Baek JeongHun 7b938741bd . 2019-07-24 16:20:25 +00:00
Baek JeongHun 5d4ed385f4
Merge pull request #36 from YacobBY/patch-1 
Nvidia Apex for FP16 calculations
 2019-07-25 00:35:01 +09:00
Yacob 2d45ba2fab
use amp grad clipping 2019-07-24 14:42:25 +02:00
Yacob a0f43fa0c0
Nvidia Apex for FP16 calculations 
Included Compatibility with the Nvidia's Apex library, which can do Floating Point16 calculations. This gives significant speedup in training. This code has been tested on a single RTX2070. If  the Nvidia Apex library is not found the code should run as normal. 

To install Apex: https://github.com/NVIDIA/apex#quick-start

Known bugs: 
-Does not work with adam parameter
-Gradient overflow keeps happening at the start, however it automatically reduces loss scale to 8192 after which this notification disappears

examples:
Loading: https://i.imgur.com/3nZROJz.png
Training: https://i.imgur.com/Q2w52m7.png
 2019-07-23 15:51:26 +02:00
Baek JeongHun 40a4100df6 update data_filtering 2019-07-17 05:41:34 +00:00
Baek JeongHun f551679a0a update data_filtering 2019-07-17 05:34:10 +00:00
Baek JeongHun ea3e322de9
Update README.md 2019-07-16 22:28:08 +09:00
Baek JeongHun dd18436360 --data_filtering_off 2019-07-16 10:04:20 +00:00
Baek JeongHun c3e07b538a fix minor 2019-07-16 03:30:44 +00:00
Baek JeongHun f15051a493 add data_filtering mode 2019-07-15 05:33:37 +00:00
Baek JeongHun 3953e51381 fix division by zero in normalized edit distance 2019-07-01 12:08:07 +00:00
Baek JeongHun f0b17556b8
Update README.md 2019-06-25 12:51:46 +09:00
Baek JeongHun 553b3689a3
Update README.md 2019-06-25 12:50:33 +09:00
Baek JeongHun 5bc60b17ca
Update README.md 2019-06-24 19:24:49 +09:00
Baek JeongHun fc59eb5eea
Update README.md 2019-06-20 14:17:01 +09:00
Baek JeongHun c076f6908f
Update README.md 2019-06-20 14:15:07 +09:00
Baek JeongHun 7c2647de8b
Update README.md 2019-06-12 05:00:16 +09:00
Baek JeongHun b5dc8fd123
Update README.md 2019-06-12 04:56:57 +09:00
Baek JeongHun bbc41a083d
Update README.md 2019-05-29 13:13:18 +09:00
Baek JeongHun 8de2cc5c32
Update README.md 2019-05-28 17:38:50 +09:00
Baek JeongHun 0a02b35df6
Update README.md 2019-05-28 17:38:22 +09:00
Baek JeongHun 16c3aadb91
Update README.md 2019-05-22 16:17:10 +09:00
Baek JeongHun 46a45478d5 update RawDataset 2019-05-21 05:47:39 +00:00
Baek JeongHun de5a153d79 add natsorted to RawDataset 2019-05-21 05:31:52 +00:00