Comments (10)
是的,这个是个很重要的对比。我们做了一些初步的实验,如下表所示。不过目前基于voxel的性能还没有调到最优,之后我们也计划开源这些baseline的代码。
from tpvformer.
另外,根据论文的说法 其实还是构造了一个[H W D C]的vox 特征空间。按照这个做法的话,内存和cpu的复杂度,貌似还是HxWxDxC。是我哪里理解不对吗?
这个因为我们只是在最后一步得到voxel的特征空间,而无需在这个voxel上进行很多操作,所以复杂度会比较低。
from tpvformer.
@wzzheng 论文还会更新voxel和tpv在显存,速度,性能上的详细对比吗?还有tpv没有实现temporal fusion,这个还会做吗?
都会更新的,目前还在测试中。temporal fusion也正在做
from tpvformer.
这个确实是一个很重要的点。作者可以做一下对比实验,获取一下数据,造福我们这些没有V100的人。。
from tpvformer.
另外,根据论文的说法
其实还是构造了一个[H W D C]的vox 特征空间。按照这个做法的话,内存和cpu的复杂度,貌似还是HxWxDxC。是我哪里理解不对吗?
from tpvformer.
这篇论文想法真的很好,我私下尝试过直接在voxel上做监督训练,但是很难训练。希望作者以后能开源用voxel训练的pipeline。
from tpvformer.
这篇论文想法真的很好,我私下尝试过直接在voxel上做监督训练,但是很难训练。希望作者以后能开源用voxel训练的pipeline。
谢谢,基于voxel的代码我们会在这两天开源,敬请期待~
from tpvformer.
@wzzheng 论文还会更新voxel和tpv在显存,速度,性能上的详细对比吗?还有tpv没有实现temporal fusion,这个还会做吗?
from tpvformer.
对于没有A100和3090的情况,如果想在Tesla V100上做tpv的相关训练测试,是否有可能呢?tpv中是否有某些设计可以修改来减小显存占用来让我们能够使用V100呢?
from tpvformer.
You can use smaller image backbone, i.e. ResNet50, smaller input image resolution, i.e. 450x800, and smaller TPV feature dimension, i.e. 64, to reduce GPU memory consumption.
We have noticed that the image backbone and the 3D reconstruction from TPV planes are the two parts that account for most of the GPU memory consumption.
from tpvformer.
Related Issues (20)
- Question about 3D OCC task training
- Minimum computer configuration for inference stage? HOT 1
- FileNotFoundError:
- What is config for the released tpv10_lidarseg_v2.pth?
- Why the visualization for lidar segmentation is in voxel?
- nuScenes数据集排列问题 HOT 4
- Higher inference resolution HOT 1
- Scale in metric unit
- Missing keys while loading pre-trained FCOS-3D weights
- got an error result when change the point_cloud_range?
- Question about positional encoding in tpvformer04
- Discrepancy in Parameter Count and FLOPs in Paper
- 为什么推理时还要读lidar bin文件?
- what is mean "Video Context"??
- Pure visual perception? HOT 1
- Visual six perspective angle calculation
- Using different H and W resolution
- How can I use with custom dataset
- error bash launcher.sh config/tpv_lidarseg.py out/tpv_lidarseg
- python visualization/vis_frame.py error
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 tpvformer.