svip-lab LBYLNet issues - Githubissues

svip-lab / LBYLNet

[CVPR2021] Look before you leap: learning landmark features for one-stage visual grounding.

47 stars 9 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

The reproduced result is lower than expected.

#13 hssdsb opened 2 weeks ago
0
DarkNet53-pretrain

#12 OliverHuang1220 closed 2 years ago
2
DarkNet53-pretrain

#11 OliverHuang1220 closed 2 years ago
1
download link of the "ReferitGame" dataset is no longer available

#10 Rose-bud opened 2 years ago
0
The reproduced performances on unc/unc+

#9 OliverHuang1220 opened 2 years ago
19
Trained models pickle files corrupted?

#8 dreichCSL closed 3 years ago
1
Does the pretrained visual backbone exclude val/test images on refcoco/+/g datasets ?

#7 seanzhuh closed 2 years ago
4
The forward( ) and backward( ) about TopLeftPool

#6 OliverHuang1220 closed 3 years ago
1
torch and Cuda's version

#5 OliverHuang1220 closed 3 years ago
1
Is BertEncoder fine-tuned on the visual grounding tasks?

#4 PlumedSerpent closed 3 years ago
2
How can I get the landmarkconv module?

#3 shydyl closed 3 years ago
0
Training getting stopped after sometime

#2 abhinavkaul95 closed 3 years ago
2
No module named 'landmarks'

#1 abhinavkaul95 closed 3 years ago
6