JingyuLi comments

Results 15 comments of


                                            JingyuLi

论文中提到的门控注意力如何实现

> 看了您的论文，非常感谢您优秀的工作。我对于文中提到的门控注意力很有兴趣，想请教您是如何实现的。在代码中没有看到门控注意力是的，我同样有这个疑问，你有实现吗？

apollo datasets

Thanks for your replay! Thanks again！or you can update the val list. goode luck!

> 嗯有意思。。。有空写多几个支持的结构。其实也不难嗯，感谢认可，yolov3这块比较熟悉，剪枝压缩这块代码原理还不太懂，修改了部分代码，现在有一个问题，对于panet的pan结构，#121 [route] layers=1 #122 [reorg3d] stride=2 #123 [route] layers=5,-1 #124 [reorg3d] stride=2 #125 [route] layers=12,-1 #126 [reorg3d] stride=2 #127 [route] layers=37,-1 #128 [reorg3d] stride=2 #129 [route] layers=62,-1 在...

Performance is lower than the results you report.

> By using the default config you provide, I have achieved 79.206 % (your result is 79.916 %) based on deeplabv3+res101 on the pascalvoc validation set. Can you tell me...

How to automatically generate ASG.

Same as you. Hello, shiche. Thanks for your good work! In supplementary materials A related to ASG. You build a 3 types relationship classifier: no relationship between two objects, subject-to-object...

Would you release the code for the pseudo label generation and assignment?

Hi, Yiwu. Same question! In my understanding, the quality of pseudo labels may greatly affect the quality of scene graph generation. Would you release the code for the pseudo label...

Would you release the code for the pseudo label generation and assignment?

> Hi @Scarecrow0 @JingyuLi-code @zyong812, thanks for your interests in our work! > > Sorry for late response. I haven't got a chance to fully clean my preprocessing code. As...

consistency_graph.pkl

We have python=3.9 torch=1.12.1+cu113 mmcv=1.6.0 mmdet=2.24.0

msvd feature

Can you download the relevant msvd_dataset feature? Can you share the link?

Swin Transformer pre-trained Model?

Thanks! Actually, I want to know the result of using "Regular ImageNet-1K trained models" and "ImageNet-22K pre-trained models" . Because for region feature and grid feature are pre-trained on Regular...