291cda7391
Update calibration results
2022-02-22 15:47:27 +01:00
9ed0c264f5
Add wandb
2022-02-22 13:36:13 +01:00
3da66347e7
Add wandb, more args, increase learning rate
2022-02-22 13:35:50 +01:00
168516924e
general improvements
2022-02-22 13:35:24 +01:00
7633990c81
Add wandb, make batch and image size configurable, fix some bugs
2022-02-22 13:32:50 +01:00
e3303cf9d4
Add wandb, make batch and image size configurable, fix some bugs, improve supervised loss function, make use of some RL stuff just implemented
2022-02-22 13:32:28 +01:00
b7dbc59c25
Add wandb, make batch and image size configurable, fix some bugs, improve supervised loss function
2022-02-22 13:31:30 +01:00
cebf776714
Add real world dataset
2022-02-22 13:29:54 +01:00
bcb1e2c6ff
Get image size from settings
2022-02-22 13:29:30 +01:00
5202ddfbe9
Add options for batch size, double heads and irl data
2022-02-22 13:28:09 +01:00
db0a8523e5
Remove useless stuf
2022-01-31 18:56:37 +01:00
28ee9003a6
Remove useless stuf
2022-01-31 18:56:18 +01:00
3f947c9dd2
WIP, buncha stuff that I didn't commit when I wrote it and now I'm afraid of losing something
2022-01-31 18:12:35 +01:00
a673f807c5
Use pytorch photometric loss instead of old self-rolled
2021-12-13 17:34:53 +01:00
f193f8d601
Remove unnecessary code from torchext
2021-12-13 17:34:28 +01:00
efac1b4dea
Update calibration data and sampled pattern
2021-12-13 17:32:48 +01:00
e0450a57b9
Fix incorrect parenthesis
2021-12-08 09:47:42 +01:00
9325503b51
use InstanceNorm2d instead of LayerNorm
2021-12-07 17:33:37 +01:00
1e91784a91
update procam calibration data and sampled pattern
2021-12-07 17:33:10 +01:00
eb8b31623a
overwrite last found datapoint
...
Since the creation crashes, the data of the last point is often incomplete and causes errors later down the pipeline. Overwriting this should alleviate this issue at a reasonable cost.
2021-11-30 13:00:26 +01:00
e7a1b59787
fix parenthesis
2021-11-30 12:56:44 +01:00
0d18f71c6d
allow pickle, otherwise we can't load all data
2021-11-30 12:52:42 +01:00
d63fedd297
fix off-by-one error
2021-11-29 18:20:39 +01:00
2d26f7a9d0
update sampled kinect pattern
2021-11-29 18:05:30 +01:00
8d7890679b
update calibration data
2021-11-29 18:03:40 +01:00
7060bd5555
add LayerNorm to conv() for testing
2021-11-29 17:54:45 +01:00
88e4a7e317
add xmltodict
2021-11-29 17:54:02 +01:00
b250ab4814
update paths
2021-11-29 17:53:44 +01:00
506a84e2d0
add .gitignore
2021-11-29 17:53:09 +01:00
119110e125
add sampled pattern to repo
2021-11-16 18:50:24 +01:00
ba7d63a7c8
Add code for rectification
...
very much wip and not working right as is
2021-11-16 18:49:48 +01:00
83d202457a
Load camera parameters from calibration xml
2021-11-15 17:29:00 +01:00
0c8529687b
Add procam calibration data
...
might require different handling once we start evaluating different tilts
2021-11-15 17:28:31 +01:00
43df77fb9b
Reformat $EVERYTHING
2021-11-15 16:53:30 +01:00
56f2aa7d5d
Add camera parameters
2021-11-15 16:52:40 +01:00
42599ee465
Add support for starting syn data generation from specific point
2021-11-15 16:50:35 +01:00
Yiyi Liao
4617525f61
added validation set and pre-trained model; tested on pytorch1.8
2021-06-07 17:54:54 +02:00
Yiyi Liao
bdcc50a88f
fix
2019-07-29 12:03:09 +02:00
Yiyi Liao
14610a160b
Merge branch 'master' of https://github.com/autonomousvision/connecting_the_dots
2019-07-29 10:40:18 +02:00
Yiyi Liao
9ce06a8a0e
fix
2019-07-29 10:39:38 +02:00
Yiyi Liao
d4d5998fe6
Merge pull request #1 from barnjamin/patch-1
...
Update readme.md
2019-06-25 14:27:15 +02:00
Ben Guidarelli
236d6bbb8f
Update readme.md
2019-06-25 06:13:50 -04:00
Yiyi Liao
f5e5c4bd3f
init
2019-06-13 16:25:11 +02:00
Yiyi Liao
26157cbb80
Initial commit
2019-06-11 11:55:18 +02:00
