Comments (9)
Not really, we only use the current frame with extra residual images.
If n_input_scans = 2, it will be (x, y, z, r, e, residual1, residual2), which you could find in our paper, section III.C.
from lidar-mos.
Yes ,I think I got the meaning of the Section III. C.
So under the e Section III. C. If I use n_input_scans=2
I just need to change the arcg_cfg.yaml , and the code in"class SemanticKitti(Dataset)", like below?
(line 4 and 10 or line 5? ) .
from lidar-mos.
Dear author, I thought I have misunderstanding the code and code comments in lines 230,2301 234 235 236 &332. ( under the sequences of n_input_scans= 2)
in line 332, there is a "concat' operation, so If there is a looping in Line 236.
the dim of 'pro_full' will be added with( x, y,z, r,e, x,y,z,r,e, ....). ? right ?
from lidar-mos.
Dear author, I thought I have misunderstanding the code and code comments in lines 230,2301 234 235 236 &332. ( under the sequences of n_input_scans= 2)
in line 332, there is a "concat' operation, so If there is a looping in Line 236. the dim of 'pro_full' will be added with( x, y,z, r,e, x,y,z,r,e, ....). ? right ?
So under the sequences of n_input_scans= 2.
under my understanding of the paper Section III.C , I think I will use the line 230 + line 235? but I got the data flow about the "proj_full". it's dims is confused.
from lidar-mos.
In our method, we only use residual images but not full multiple scans.
We've however also tried to directly concatenate multiple scans together as you mentioned 230+235 and didn't get much improvement.
You could give it a try, and I only test it for two scans for an ablation which could be buggy for more scans.
I hope my answer helps.
from lidar-mos.
In our method, we only use residual images but not full multiple scans. We've however also tried to directly concatenate multiple scans together as you mentioned 230+235 and didn't get much improvement.
You could give it a try, and I only test it for two scans for an ablation which could be buggy for more scans.
I hope my answer helps.
Dear author,
Maybe I misunderstand the meanning of the " multi scan and multi residuals images"
I just want to follow the work in the section III.C with use multi residual images.
So now If I just have the two residual images ,residual_1,residual_2, just look at the image, like the code in line 243.
I dont need to change the Line 231& 236, just only change the n_inupt_scans to 2 in arch_cfg.yaml, ?
If it is right ,I think I have understanding the meaning.!!!
thanks for bother you again.
from lidar-mos.
Yes, if you only want to use residual images as what we used in our paper, you need then just change the config file.
可能中文沟通更容易一点,如果你只是想跟我们一样用当前帧的(x, y, z, r, e)加residual images。你只需要修改配置文件里面的n_input_scans就行了。
如果你想用多帧激光的信息(x1, y1, z1, r1, e1, x2, ..., e2)你可以用我们之前测试的代码230 + 235,但是这不是我所提的方法,我只测试过两帧的,对于多帧的情况可能会有问题。
希望我的回答是清楚的,如果你觉得有帮助,请帮我点个赞,谢谢!
from lidar-mos.
谢谢您,终于理解了,finally,
thanks!!!!
from lidar-mos.
请问一下,对于每帧生成8张残差图像的代码还没有开源是吗?我自己改了一下代码,但是和论文中8张残差图像的效果不太一样。
from lidar-mos.
Related Issues (20)
- How to train this code on multi-GPUs? HOT 4
- How to re-train when training is unexpectedly interrupted? HOT 8
- How to use the pretrained model to test my own Lidar Scans? HOT 5
- About set my own Lidar pose to kitti format pose HOT 3
- how to visualize? HOT 4
- How did you define the moving and static object HOT 4
- About using my own lidar data and lidar pose HOT 2
- Salsanext HOT 20
- 关于使用rangenet的问题 (Problem when using rangenet) HOT 4
- deployment
- Headless way to visualize the result? HOT 1
- About training time HOT 3
- moving and static object
- 在保存训练模型时遇到问题 (Problem while saving training model)
- Try with different sensor
- salsanext issue
- 请问如何对预训练模型进行微调?
- core dumped when infering with SalsaNext
- About the issue of multi-GPU training.
- ubuntu or windows HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lidar-mos.