amingwu / single-dgod Goto Github PK
View Code? Open in Web Editor NEWLicense: MIT License
License: MIT License
老师您好,看完您的论文之后,对论文解纠缠部分的方法很感兴趣,希望可以分享一下论文完整代码!非常感谢!
我的邮箱是[email protected]
Can you help me solve the following error when I run the trainval_net_fpn.py? Thank you very much!
python trainval_net_fpn.py --dataset dc_fpn --net res101 --epochs 20 --bs 2 --nw 8 --lr 0.004 --lr_decay_step 8 --cuda
Traceback (most recent call last):
File "/home/pxs/Single-DGOD/trainval_net_fpn.py", line 36, in
from model.faster_rcnn.resnet_fpn import resnet
File "/home/pxs/Single-DGOD/lib/model/faster_rcnn/resnet_fpn.py", line 6, in
from model.faster_rcnn.fpn import _FPN
File "/home/pxs/Single-DGOD/lib/model/faster_rcnn/fpn.py", line 13, in
from model.rpn_fpn.rpn_fpn import _RPN_FPN
File "/home/pxs/Single-DGOD/lib/model/rpn_fpn/rpn_fpn.py", line 7, in
from .proposal_layer_fpn import _ProposalLayer_FPN
File "/home/pxs/Single-DGOD/lib/model/rpn_fpn/proposal_layer_fpn.py", line 21, in
from model.roi_layers import nms
File "/home/pxs/Single-DGOD/lib/model/roi_layers/init.py", line 3, in
from .nms import nms
File "/home/pxs/Single-DGOD/lib/model/roi_layers/nms.py", line 3, in
from model import _C
ImportError: cannot import name '_C' from 'model' (/home/pxs/Single-DGOD/lib/model/init.py)
作者您好,这里好像不是您作品完整的代码,我对您作品中的循环解耦很感兴趣,请问您可以分享一下Single DGOD详细的代码嘛,十分感谢!期待您的回复!
Can you help me solve the following error when I run the trainval_net_fpn.py? Thank you very much!
Thanks for your guidance!
这份代码是不是少了一个model的文件???
python setup.py install always fail
您好,请问方便分享下解纠缠的相关代码吗?
我的邮箱是:[email protected]
感谢!
We want to gain a better understanding of your work, when will you release the code?
Thank you very much for your work. When will you open source code?
@AmingWu Thanks for putting some of the code up!
From my understanding, this repository is built upon some existing code base. I see a lot of imports that suggest so, but I'd love to know what the exact codebase used for this is / get a more elaborate requirements.txt
Thanks!
作者您好!对您的工作很感兴趣,能麻烦发一下完整的源码吗,十分感谢!
Email:[email protected]
Hi Aming Wu,
Thank you for sharing the code. I am trying to reproduce the results mentioned in your paper but currently I am unable to locate your method in the code e.g the self-distillation loss. Is the shared code only contains Faster RCNN implantation?
Can you talk about how the method comparison in Section 4.2 inserted these methods into Faster R-CNN?
Can you talk about how the method comparison in Section 4.2 inserted these methods into Faster R-CNN?In addition, where can I find the supplementary materials mentioned in the implementation details?
您好,看完论文之后,对论文方法很感兴趣,但GitHub上面的代码缺少循环解纠缠,可以分享一下论文完整代码吗?谢谢!
我的邮箱是[email protected]
Thanks. Please contact me ([email protected]). I will give you the code of the disentangle. Welcome to contact me.
Thanks a lot
Your code is just a baseline of faster-RCNN with FPN, can you share the whole code especially the Cyclic-Distentangled part.
武老师好!
在原论文中,我没有看到您提出方法Single-DGOD 及Faster rcnn的输入图片尺寸,请问老师可以说明这两种算法的输入图片尺寸吗?谢谢!
There is no test.txt in Dusk-rainy.zip and Night-rainy.zip,
How can I do eval on these datasets
作者您好:
最近拜读了您发表在CVPR2022上面的这篇论文,相关机制对我深受启发,因此想获得完整源码作进一步学习了解。
我的邮箱是:[email protected],感谢!
@AmingWu Hello, I want to know the specific meaning of the mAP in the experimental results. Is it mAP50 or mAP50-95?
can anyone please tell me how do i actually reproduce the code?
First of all, thank you very much for setting set by the author
I have completely reproduced many baseline by using the author's data set
The results are almost consistent with the author's description, which makes me feel happy and admires the author's scientific research level
The author's many contributions will be beneficial to the development of the community
But there are still some problems in the paper, which I don't quite understand、
As a freshman, the problem may be a little low-level. I hope the author can be a little tolerant
Q:
1.feature normalization methods are a litte less than baseline(faster-rcnn) Is there no need for comparison?
2. In Ablation Analysis,Why do test on the training set?(Daytime-Sunny)
3. Decoupling in previous articles of the author usually takes the way of updating parameters step by step,however,In this article, an updated way is adopted as same as DDF(Decompose to Adapt: Cross-domain Object Detection via Feature Disentanglement).
In my view,according to experiment which only one data set is trained, and then the expectation and variance on convolution between different data sets are observed。I found that on two data sets with domain shift, the trained backbone was not consistent with the data distribution of the following layers。In the nutshell, shallow decoupling information will still introduce domain shift after some convolution?
Is the next development of decoupling method to solve the problem of preventing shallow information from introducing domain shift?
Last but no least,thanks to the author for his contribution to the domain adaptive community and his pioneering approach to feature decoupling
拜读完您的论文,想研究一下代码,请问可以分享一下完整代码吗?十分感谢,如果您有时间烦请发送至邮箱:
[email protected]
Could you share the experimental results of single-DGOD on FPN structure?
README only provides experimental results of other 5 methods
武老师好,
在阅读您的Single-Domain Generalized Object Detection工作时,需要结合完整源码学习,请问老师可以方便提供吗?我的邮箱是[email protected]
请求论文完整代码,谢谢!
我的邮箱是[email protected]
您好,可以分享一下循环解纠缠的详细代码吗?谢谢
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.